{"cells":[{"metadata":{},"cell_type":"markdown","source":"# CISC 351 Group 3 - Google Analytics Customer Revenue Prediction"},{"metadata":{"_kg_hide-input":false},"cell_type":"markdown","source":"In order to perform this regression task, the overall process followed is shown in the contents below. This process started off with the importation of relevant libraries and a prelimary assesment of the data. Afterwards, tasks such as exploratory analysis and the preprocessing of data was conducted. Lastly, the 2 baseline models were built, implemented, and evaluated. Detailed discussion can be found in the report handed in.\n\n1. [Preliminary Data Assesment](#1)\n2. [Preprocessing Data](#2)\n 1. [JSon Parser Function](#2.1)\n 2. [Function to load in a Random 1% Sample of Data](#2.2)\n 3. [Exploratory Analysis](#2.3) \n 4. [Feature Engineering](#2.4) \n3. [Building 2 Baseline Models](#3)\n 1. [Separate Features, Target,and Construct Validation Set](#3.1) \n 2. [LGBM](#3.2)\n 3. [XGBoost](#3.3)\n 4. [Summary and Comparison of the Two Models](#3.4) "},{"metadata":{},"cell_type":"markdown","source":"\n\n## Preliminary Data Assesment"},{"metadata":{},"cell_type":"markdown","source":"With any machine learning task, the team started off with importing the necessary libraries to perform the regression task. A portion of the data was read in and assessed. In addition, the sizes of the training and test sets was analyzed. This analysis found two difficulties in dealing with the dataset. First, the data was parsed in JSon queries. Second, the dataset size was very large, with the training data having over a million rows and the testing data having over several hundred thousand rows. Discussions on how the team dealt with this issue take place in the following sections."},{"metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","trusted":true},"cell_type":"code","source":"import os\nimport json\nimport numpy as np\nimport pandas as pd\nfrom pandas.io.json import json_normalize\nimport random\nimport matplotlib.pyplot as plt\nimport multiprocessing as mp\nimport datetime as dt\nimport xgboost as xgb\nimport sklearn\nimport lightgbm as lgb\nfrom scipy.stats import randint as sp_randint\nfrom scipy.stats import uniform as sp_uniform\nfrom sklearn.model_selection import RandomizedSearchCV, GridSearchCV\nimport time\nimport sklearn.metrics\nimport plotly.plotly as py\nimport plotly.graph_objs as go\nfrom plotly import tools\nfrom plotly.offline import init_notebook_mode, iplot","execution_count":37,"outputs":[]},{"metadata":{"_uuid":"4f8c7056484b0ddc504c588b27292f24a93c1451","trusted":true},"cell_type":"code","source":"#Determine Sizes of Training and Testing Sets\ntraining_size = int(sum(1 for line in open('../input/train_v2.csv')))\ntesting_size = int(sum(1 for line in open('../input/test_v2.csv')))\nprint('Size of training set is: ', training_size, 'rows. Size of testing set is: ', testing_size, 'rows.')","execution_count":38,"outputs":[{"output_type":"stream","text":"Size of training set is: 1708338 rows. Size of testing set is: 401590 rows.\n","name":"stdout"}]},{"metadata":{"_uuid":"6b3ec34eca8a79337bf45ae9ac597727ed6f4f1e","scrolled":true,"trusted":true},"cell_type":"code","source":"#Load portion of training data to assess the data\nJSON_COLUMNS = ['device', 'geoNetwork', 'totals', 'trafficSource']\ndf2 = pd.read_csv('../input/train_v2.csv', nrows=1000)\ndf2.head()","execution_count":39,"outputs":[{"output_type":"execute_result","execution_count":39,"data":{"text/plain":" channelGrouping ... visitStartTime\n0 Organic Search ... 1508198450\n1 Referral ... 1508176307\n2 Direct ... 1508201613\n3 Organic Search ... 1508169851\n4 Organic Search ... 1508190552\n\n[5 rows x 13 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
channelGroupingcustomDimensionsdatedevicefullVisitorIdgeoNetworkhitssocialEngagementTypetotalstrafficSourcevisitIdvisitNumbervisitStartTime
0Organic Search[{'index': '4', 'value': 'EMEA'}]20171016{\"browser\": \"Firefox\", \"browserVersion\": \"not ...3162355547410993243{\"continent\": \"Europe\", \"subContinent\": \"Weste...[{'hitNumber': '1', 'time': '0', 'hour': '17',...Not Socially Engaged{\"visits\": \"1\", \"hits\": \"1\", \"pageviews\": \"1\",...{\"campaign\": \"(not set)\", \"source\": \"google\", ...150819845011508198450
1Referral[{'index': '4', 'value': 'North America'}]20171016{\"browser\": \"Chrome\", \"browserVersion\": \"not a...8934116514970143966{\"continent\": \"Americas\", \"subContinent\": \"Nor...[{'hitNumber': '1', 'time': '0', 'hour': '10',...Not Socially Engaged{\"visits\": \"1\", \"hits\": \"2\", \"pageviews\": \"2\",...{\"referralPath\": \"/a/google.com/transportation...150817630761508176307
2Direct[{'index': '4', 'value': 'North America'}]20171016{\"browser\": \"Chrome\", \"browserVersion\": \"not a...7992466427990357681{\"continent\": \"Americas\", \"subContinent\": \"Nor...[{'hitNumber': '1', 'time': '0', 'hour': '17',...Not Socially Engaged{\"visits\": \"1\", \"hits\": \"2\", \"pageviews\": \"2\",...{\"campaign\": \"(not set)\", \"source\": \"(direct)\"...150820161311508201613
3Organic Search[{'index': '4', 'value': 'EMEA'}]20171016{\"browser\": \"Chrome\", \"browserVersion\": \"not a...9075655783635761930{\"continent\": \"Asia\", \"subContinent\": \"Western...[{'hitNumber': '1', 'time': '0', 'hour': '9', ...Not Socially Engaged{\"visits\": \"1\", \"hits\": \"2\", \"pageviews\": \"2\",...{\"campaign\": \"(not set)\", \"source\": \"google\", ...150816985111508169851
4Organic Search[{'index': '4', 'value': 'Central America'}]20171016{\"browser\": \"Chrome\", \"browserVersion\": \"not a...6960673291025684308{\"continent\": \"Americas\", \"subContinent\": \"Cen...[{'hitNumber': '1', 'time': '0', 'hour': '14',...Not Socially Engaged{\"visits\": \"1\", \"hits\": \"2\", \"pageviews\": \"2\",...{\"campaign\": \"(not set)\", \"source\": \"google\", ...150819055211508190552
\n
"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"\n## Preprocessing Data"},{"metadata":{},"cell_type":"markdown","source":"\n### JSon Parser Function"},{"metadata":{},"cell_type":"markdown","source":"As described above, one of the first issues encountered with the dataset was that it was parsed in JSon queries. After trial and error, the the following function was defined succesfully to parse the JSon queries in Pandas dataframes. "},{"metadata":{"_uuid":"ca110b71b299e13fb9859ba41b2ed4e320d1b36c","trusted":true},"cell_type":"code","source":"#Define function to parse the JSON entries\ndef Parser (df):\n 'This function takes a Pandas dataframe as an input, parses the JSON columns in the dataframe, and produces a parsed Pandas dataframe'\n JSON_COLUMNS = ['device', 'geoNetwork', 'totals', 'trafficSource']\n for column in JSON_COLUMNS:\n column_as_df = json_normalize(df[column])\n column_as_df.columns = [f\"{column}.{subcolumn}\" for subcolumn in column_as_df.columns]\n df = df.drop(column, axis=1).merge(column_as_df, right_index=True, left_index=True)\n return df","execution_count":40,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"\n### Function to load in a Random 1% Sample of Data"},{"metadata":{},"cell_type":"markdown","source":"As described in the preliminary data assesment, there were two approaches taken to dealing with the massive size of the dataset. The first approach taken was to load and preprocess the data in batches with a function called the dataPipeline function that can be found commented out further down in the code. The idea with this function was to load the data in 1% batches, preprocess each batch thereby reducing the dataframe size signficantly enough that once all the 1% batches were concatenated together, the dataframe would be able to fit into memory. The function ran and produced an dataframe output, but the result was not to expectation. Due to a time constraint on the project, the team decided to use the function defined below called load_DF that generates a random 1% sample of the training and test set. This function is initialized with a random generator with random.seed(25) to produce the same random dataframe everytime. As this function produces a completely random dataframe, the dataframe distribution should be representative of the entire dataframe."},{"metadata":{"_uuid":"749dae84b9f202c9d1bec45cc7a31e1683fb0ab3","trusted":true},"cell_type":"code","source":"# Define function to randomly load in 1% of train/test data\ndef load_DF(filename):\n 'This function takes a csv file as an input, and outputs 1% of that files rows randomly selected '\n JSON_COLUMNS = ['device', 'geoNetwork', 'totals', 'trafficSource'] #JSon Columns in the training set\n num_lines = int(sum(1 for line in open(filename))) # Read number of file lines in the training dataset\n print('The number of rows in the file is ',num_lines) \n n = int(num_lines-1) #number of records in file (excludes header)\n s = int(0.01*n) #Desired sample of 10 Percent of Dataset\n random.seed(25)\n skip = sorted(random.sample(range(1,n+1),n-s)) #the 0-indexed header will not be included in the skip list\n DF=pd.read_csv(filename, converters={column: json.loads for column in JSON_COLUMNS}, \n dtype={'fullVisitorId': 'str'},skiprows=skip)\n \n DF = Parser(DF)\n return DF","execution_count":41,"outputs":[]},{"metadata":{"_uuid":"5aecf331192c206658bc1227e0ab0d8d9392f97d","trusted":true},"cell_type":"code","source":"#Unparse Randomly chosen training data for feature engineering\ntrainingData = load_DF('../input/train_v2.csv')\n#Unparse Randomly chosen testing data for feature engineering\ntestingData = load_DF('../input/test_v2.csv')","execution_count":42,"outputs":[{"output_type":"stream","text":"The number of rows in the file is 1708338\nThe number of rows in the file is 401590\n","name":"stdout"}]},{"metadata":{"_uuid":"a5cbd22b749d3916e3fd9a75bd59c031cd7cd0a2","trusted":true},"cell_type":"code","source":"print(trainingData.shape)\ntrainingData.head()","execution_count":43,"outputs":[{"output_type":"stream","text":"(17083, 59)\n","name":"stdout"},{"output_type":"execute_result","execution_count":43,"data":{"text/plain":" channelGrouping ... trafficSource.source\n0 Direct ... (direct)\n1 Organic Search ... google\n2 Affiliates ... Partners\n3 Organic Search ... bing\n4 Organic Search ... google\n\n[5 rows x 59 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
channelGroupingcustomDimensionsdatefullVisitorIdhitssocialEngagementTypevisitIdvisitNumbervisitStartTimedevice.browserdevice.browserSizedevice.browserVersiondevice.deviceCategorydevice.flashVersiondevice.isMobiledevice.languagedevice.mobileDeviceBrandingdevice.mobileDeviceInfodevice.mobileDeviceMarketingNamedevice.mobileDeviceModeldevice.mobileInputSelectordevice.operatingSystemdevice.operatingSystemVersiondevice.screenColorsdevice.screenResolutiongeoNetwork.citygeoNetwork.cityIdgeoNetwork.continentgeoNetwork.countrygeoNetwork.latitudegeoNetwork.longitudegeoNetwork.metrogeoNetwork.networkDomaingeoNetwork.networkLocationgeoNetwork.regiongeoNetwork.subContinenttotals.bouncestotals.hitstotals.newVisitstotals.pageviewstotals.sessionQualityDimtotals.timeOnSitetotals.totalTransactionRevenuetotals.transactionRevenuetotals.transactionstotals.visitstrafficSource.adContenttrafficSource.adwordsClickInfo.adNetworkTypetrafficSource.adwordsClickInfo.criteriaParameterstrafficSource.adwordsClickInfo.gclIdtrafficSource.adwordsClickInfo.isVideoAdtrafficSource.adwordsClickInfo.pagetrafficSource.adwordsClickInfo.slottrafficSource.campaigntrafficSource.isTrueDirecttrafficSource.keywordtrafficSource.mediumtrafficSource.referralPathtrafficSource.source
0Direct[{'index': '4', 'value': 'North America'}]201710164447569759370230345[{'hitNumber': '1', 'time': '0', 'hour': '18',...Not Socially Engaged150820327911508203279Chromenot available in demo datasetnot available in demo datasetmobilenot available in demo datasetTruenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetAndroidnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetnot available in demo datasetpsu.edunot available in demo datasetnot available in demo datasetNorthern AmericaNaN414166NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaN(not set)TrueNaN(none)NaN(direct)
1Organic Search[{'index': '4', 'value': 'North America'}]201710168449727979333841797[{'hitNumber': '1', 'time': '0', 'hour': '15',...Not Socially Engaged150819259711508192597Chromenot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetLinuxnot available in demo datasetnot available in demo datasetnot available in demo datasetSunnyvalenot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetSan Francisco-Oakland-San Jose CA(not set)not available in demo datasetCaliforniaNorthern AmericaNaN616240NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaN(not set)NaN(not provided)organicNaNgoogle
2Affiliates[{'index': '4', 'value': 'EMEA'}]201710160258560966793548223[{'hitNumber': '1', 'time': '0', 'hour': '7', ...Not Socially Engaged150816375511508163755Internet Explorernot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetWindowsnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetEuropeNetherlandsnot available in demo datasetnot available in demo datasetnot available in demo datasetkpn.netnot available in demo datasetnot available in demo datasetWestern EuropeNaN7171243NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaNData Share PromoNaNNaNaffiliateNaNPartners
3Organic Search[{'index': '4', 'value': 'Central America'}]201710162120129608233101123[{'hitNumber': '1', 'time': '0', 'hour': '11',...Not Socially Engaged150817726111508177261Internet Explorernot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetWindowsnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetAmericasMexiconot available in demo datasetnot available in demo datasetnot available in demo datasetunknown.unknownnot available in demo datasetnot available in demo datasetCentral AmericaNaN8181111NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaN(not set)NaNgoogle online merchandiseorganicNaNbing
4Organic Search[{'index': '4', 'value': 'North America'}]201710163848523864758719206[{'hitNumber': '1', 'time': '0', 'hour': '14',...Not Socially Engaged150818899511508188995Chromenot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetChrome OSnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetnot available in demo dataset(not set)not available in demo datasetnot available in demo datasetNorthern AmericaNaN9176250NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaN(not set)NaN(not provided)organicNaNgoogle
\n
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"trainingData.info()","execution_count":44,"outputs":[{"output_type":"stream","text":"\nRangeIndex: 17083 entries, 0 to 17082\nData columns (total 59 columns):\nchannelGrouping 17083 non-null object\ncustomDimensions 17083 non-null object\ndate 17083 non-null int64\nfullVisitorId 17083 non-null object\nhits 17083 non-null object\nsocialEngagementType 17083 non-null object\nvisitId 17083 non-null int64\nvisitNumber 17083 non-null int64\nvisitStartTime 17083 non-null int64\ndevice.browser 17083 non-null object\ndevice.browserSize 17083 non-null object\ndevice.browserVersion 17083 non-null object\ndevice.deviceCategory 17083 non-null object\ndevice.flashVersion 17083 non-null object\ndevice.isMobile 17083 non-null bool\ndevice.language 17083 non-null object\ndevice.mobileDeviceBranding 17083 non-null object\ndevice.mobileDeviceInfo 17083 non-null object\ndevice.mobileDeviceMarketingName 17083 non-null object\ndevice.mobileDeviceModel 17083 non-null object\ndevice.mobileInputSelector 17083 non-null object\ndevice.operatingSystem 17083 non-null object\ndevice.operatingSystemVersion 17083 non-null object\ndevice.screenColors 17083 non-null object\ndevice.screenResolution 17083 non-null object\ngeoNetwork.city 17083 non-null object\ngeoNetwork.cityId 17083 non-null object\ngeoNetwork.continent 17083 non-null object\ngeoNetwork.country 17083 non-null object\ngeoNetwork.latitude 17083 non-null object\ngeoNetwork.longitude 17083 non-null object\ngeoNetwork.metro 17083 non-null object\ngeoNetwork.networkDomain 17083 non-null object\ngeoNetwork.networkLocation 17083 non-null object\ngeoNetwork.region 17083 non-null object\ngeoNetwork.subContinent 17083 non-null object\ntotals.bounces 8793 non-null object\ntotals.hits 17083 non-null object\ntotals.newVisits 13110 non-null object\ntotals.pageviews 17079 non-null object\ntotals.sessionQualityDim 8774 non-null object\ntotals.timeOnSite 8254 non-null object\ntotals.totalTransactionRevenue 153 non-null object\ntotals.transactionRevenue 153 non-null object\ntotals.transactions 154 non-null object\ntotals.visits 17083 non-null object\ntrafficSource.adContent 646 non-null object\ntrafficSource.adwordsClickInfo.adNetworkType 750 non-null object\ntrafficSource.adwordsClickInfo.criteriaParameters 17083 non-null object\ntrafficSource.adwordsClickInfo.gclId 751 non-null object\ntrafficSource.adwordsClickInfo.isVideoAd 750 non-null object\ntrafficSource.adwordsClickInfo.page 750 non-null object\ntrafficSource.adwordsClickInfo.slot 750 non-null object\ntrafficSource.campaign 17083 non-null object\ntrafficSource.isTrueDirect 5415 non-null object\ntrafficSource.keyword 6541 non-null object\ntrafficSource.medium 17083 non-null object\ntrafficSource.referralPath 5623 non-null object\ntrafficSource.source 17083 non-null object\ndtypes: bool(1), int64(4), object(54)\nmemory usage: 7.6+ MB\n","name":"stdout"}]},{"metadata":{"_uuid":"fd8d9837f0a332d34c3e06252ea9372bf8b8c71d","trusted":true},"cell_type":"code","source":"print(testingData.shape)\ntestingData.head()","execution_count":45,"outputs":[{"output_type":"stream","text":"(4015, 59)\n","name":"stdout"},{"output_type":"execute_result","execution_count":45,"data":{"text/plain":" channelGrouping ... trafficSource.source\n0 Referral ... (direct)\n1 Referral ... (direct)\n2 Organic Search ... google\n3 Organic Search ... google\n4 Direct ... (direct)\n\n[5 rows x 59 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
channelGroupingcustomDimensionsdatefullVisitorIdhitssocialEngagementTypevisitIdvisitNumbervisitStartTimedevice.browserdevice.browserSizedevice.browserVersiondevice.deviceCategorydevice.flashVersiondevice.isMobiledevice.languagedevice.mobileDeviceBrandingdevice.mobileDeviceInfodevice.mobileDeviceMarketingNamedevice.mobileDeviceModeldevice.mobileInputSelectordevice.operatingSystemdevice.operatingSystemVersiondevice.screenColorsdevice.screenResolutiongeoNetwork.citygeoNetwork.cityIdgeoNetwork.continentgeoNetwork.countrygeoNetwork.latitudegeoNetwork.longitudegeoNetwork.metrogeoNetwork.networkDomaingeoNetwork.networkLocationgeoNetwork.regiongeoNetwork.subContinenttotals.bouncestotals.hitstotals.newVisitstotals.pageviewstotals.sessionQualityDimtotals.timeOnSitetotals.totalTransactionRevenuetotals.transactionRevenuetotals.transactionstotals.visitstrafficSource.adContenttrafficSource.adwordsClickInfo.adNetworkTypetrafficSource.adwordsClickInfo.criteriaParameterstrafficSource.adwordsClickInfo.gclIdtrafficSource.adwordsClickInfo.isVideoAdtrafficSource.adwordsClickInfo.pagetrafficSource.adwordsClickInfo.slottrafficSource.campaigntrafficSource.isTrueDirecttrafficSource.keywordtrafficSource.mediumtrafficSource.referralPathtrafficSource.source
0Referral[{'index': '4', 'value': 'North America'}]201805115299820794845174508[{'hitNumber': '1', 'time': '0', 'hour': '18',...Not Socially Engaged152608793721526087937Chromenot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetLinuxnot available in demo datasetnot available in demo datasetnot available in demo datasetKirklandnot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetSeattle-Tacoma WA(not set)not available in demo datasetWashingtonNorthern AmericaNaN5NaN3128NaNNaNNaN1(not set)NaNnot available in demo datasetNaNNaNNaNNaN(not set)NaNNaN(none)/(direct)
1Referral[{'index': '4', 'value': 'North America'}]201805113326540121689239742[{'hitNumber': '1', 'time': '0', 'hour': '12',...Not Socially Engaged152606849111526068491Chromenot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetLinuxnot available in demo datasetnot available in demo datasetnot available in demo datasetAnn Arbornot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetDetroit MI(not set)not available in demo datasetMichiganNorthern AmericaNaN514142NaNNaNNaN1(not set)NaNnot available in demo datasetNaNNaNNaNNaN(not set)NaNNaN(none)/(direct)
2Organic Search[{'index': '4', 'value': 'North America'}]201805110121413971533176085[{'hitNumber': '1', 'time': '0', 'hour': '21',...Not Socially Engaged152609786811526097868Safarinot available in demo datasetnot available in demo datasetmobilenot available in demo datasetTruenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetiOSnot available in demo datasetnot available in demo datasetnot available in demo datasetFremontnot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetSan Francisco-Oakland-San Jose CAcomcast.netnot available in demo datasetCaliforniaNorthern AmericaNaN614160NaNNaNNaN1(not set)NaNnot available in demo datasetNaNNaNNaNNaN(not set)NaN(not provided)organic(not set)google
3Organic Search[{'index': '4', 'value': 'North America'}]201805118816603995507550690[{'hitNumber': '1', 'time': '0', 'hour': '15',...Not Socially Engaged152607758321526077583Chromenot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetChrome OSnot available in demo datasetnot available in demo datasetnot available in demo datasetSan Josenot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetSan Francisco-Oakland-San Jose CA(not set)not available in demo datasetCaliforniaNorthern AmericaNaN6NaN58258NaNNaNNaN1(not set)NaNnot available in demo datasetNaNNaNNaNNaN(not set)TrueNaNorganic(not set)google
4Direct[{'index': '4', 'value': 'North America'}]201805116216204101074901430[{'hitNumber': '1', 'time': '0', 'hour': '7', ...Not Socially Engaged152604957111526049571Chromenot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetMacintoshnot available in demo datasetnot available in demo datasetnot available in demo datasetSan Josenot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetSan Francisco-Oakland-San Jose CA(not set)not available in demo datasetCaliforniaNorthern AmericaNaN1411252387NaNNaNNaN1(not set)NaNnot available in demo datasetNaNNaNNaNNaN(not set)True(not set)(none)(not set)(direct)
\n
"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"\n### Exploratory Analysis\n"},{"metadata":{},"cell_type":"markdown","source":"#### Constant Columns & Missing Values Analysis\n\nAfter loading in the random 1% sample of the training and testing data, the team explored the data and determined the number of columns with constant values and performed a missing values analysis on the two dataframes with the missing_values function. This analysis determined that both the training and testing datasets had a large number of columns with constant values and a large amount of missing values. These will be discussed further in the analysis of the Feature Engineering function."},{"metadata":{"trusted":true},"cell_type":"code","source":"# Determine the number of columns in dataset with constant values \ncolumns_constant = [i for i in trainingData.columns if trainingData[i].nunique(dropna=False)==1 ]\ncolumns_constant","execution_count":46,"outputs":[{"output_type":"execute_result","execution_count":46,"data":{"text/plain":"['socialEngagementType',\n 'device.browserSize',\n 'device.browserVersion',\n 'device.flashVersion',\n 'device.language',\n 'device.mobileDeviceBranding',\n 'device.mobileDeviceInfo',\n 'device.mobileDeviceMarketingName',\n 'device.mobileDeviceModel',\n 'device.mobileInputSelector',\n 'device.operatingSystemVersion',\n 'device.screenColors',\n 'device.screenResolution',\n 'geoNetwork.cityId',\n 'geoNetwork.latitude',\n 'geoNetwork.longitude',\n 'geoNetwork.networkLocation',\n 'totals.visits',\n 'trafficSource.adwordsClickInfo.criteriaParameters']"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Analyze the missing values in the dataset \ndef missing_values(data):\n 'Input: Dataframe, Output: Dataframe. Function analyzes the missing values in a dataframe and returns the percentage missing.'\n #Count up all the null values \n missing = data.isnull().sum().sort_values(ascending = False) \n # Get Count of total values \n Count= data.count().sort_values(ascending=False)\n #Express Missing values as a percent of total values \n Percent = (missing/(Count+missing)*100).sort_values(ascending=False)\n df = pd.concat([missing, Count, Percent], axis=1, keys=['Missing', 'Count', 'Percent'])\n print (df[~(df['Percent'] == 0)]) # Returning values of nulls different of 0 \n return df","execution_count":47,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"missing_values(trainingData)","execution_count":48,"outputs":[{"output_type":"stream","text":" Missing ... Percent\ntotals.bounces 8290 ... 48.527776\ntotals.newVisits 3973 ... 23.257039\ntotals.pageviews 4 ... 0.023415\ntotals.sessionQualityDim 8309 ... 48.638998\ntotals.timeOnSite 8829 ... 51.682960\ntotals.totalTransactionRevenue 16930 ... 99.104373\ntotals.transactionRevenue 16930 ... 99.104373\ntotals.transactions 16929 ... 99.098519\ntrafficSource.adContent 16437 ... 96.218463\ntrafficSource.adwordsClickInfo.adNetworkType 16333 ... 95.609670\ntrafficSource.adwordsClickInfo.gclId 16332 ... 95.603817\ntrafficSource.adwordsClickInfo.isVideoAd 16333 ... 95.609670\ntrafficSource.adwordsClickInfo.page 16333 ... 95.609670\ntrafficSource.adwordsClickInfo.slot 16333 ... 95.609670\ntrafficSource.isTrueDirect 11668 ... 68.301821\ntrafficSource.keyword 10542 ... 61.710472\ntrafficSource.referralPath 11460 ... 67.084236\n\n[17 rows x 3 columns]\n","name":"stdout"},{"output_type":"stream","text":"/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:10: FutureWarning:\n\nSorting because non-concatenation axis is not aligned. A future version\nof pandas will change to not sort by default.\n\nTo accept the future behavior, pass 'sort=False'.\n\nTo retain the current behavior and silence the warning, pass 'sort=True'.\n\n\n","name":"stderr"},{"output_type":"execute_result","execution_count":48,"data":{"text/plain":" Missing ... Percent\nchannelGrouping 0 ... 0.000000\ncustomDimensions 0 ... 0.000000\ndate 0 ... 0.000000\ndevice.browser 0 ... 0.000000\ndevice.browserSize 0 ... 0.000000\ndevice.browserVersion 0 ... 0.000000\ndevice.deviceCategory 0 ... 0.000000\ndevice.flashVersion 0 ... 0.000000\ndevice.isMobile 0 ... 0.000000\ndevice.language 0 ... 0.000000\ndevice.mobileDeviceBranding 0 ... 0.000000\ndevice.mobileDeviceInfo 0 ... 0.000000\ndevice.mobileDeviceMarketingName 0 ... 0.000000\ndevice.mobileDeviceModel 0 ... 0.000000\ndevice.mobileInputSelector 0 ... 0.000000\ndevice.operatingSystem 0 ... 0.000000\ndevice.operatingSystemVersion 0 ... 0.000000\ndevice.screenColors 0 ... 0.000000\ndevice.screenResolution 0 ... 0.000000\nfullVisitorId 0 ... 0.000000\ngeoNetwork.city 0 ... 0.000000\ngeoNetwork.cityId 0 ... 0.000000\ngeoNetwork.continent 0 ... 0.000000\ngeoNetwork.country 0 ... 0.000000\ngeoNetwork.latitude 0 ... 0.000000\ngeoNetwork.longitude 0 ... 0.000000\ngeoNetwork.metro 0 ... 0.000000\ngeoNetwork.networkDomain 0 ... 0.000000\ngeoNetwork.networkLocation 0 ... 0.000000\ngeoNetwork.region 0 ... 0.000000\ngeoNetwork.subContinent 0 ... 0.000000\nhits 0 ... 0.000000\nsocialEngagementType 0 ... 0.000000\ntotals.bounces 8290 ... 48.527776\ntotals.hits 0 ... 0.000000\ntotals.newVisits 3973 ... 23.257039\ntotals.pageviews 4 ... 0.023415\ntotals.sessionQualityDim 8309 ... 48.638998\ntotals.timeOnSite 8829 ... 51.682960\ntotals.totalTransactionRevenue 16930 ... 99.104373\ntotals.transactionRevenue 16930 ... 99.104373\ntotals.transactions 16929 ... 99.098519\ntotals.visits 0 ... 0.000000\ntrafficSource.adContent 16437 ... 96.218463\ntrafficSource.adwordsClickInfo.adNetworkType 16333 ... 95.609670\ntrafficSource.adwordsClickInfo.criteriaParameters 0 ... 0.000000\ntrafficSource.adwordsClickInfo.gclId 16332 ... 95.603817\ntrafficSource.adwordsClickInfo.isVideoAd 16333 ... 95.609670\ntrafficSource.adwordsClickInfo.page 16333 ... 95.609670\ntrafficSource.adwordsClickInfo.slot 16333 ... 95.609670\ntrafficSource.campaign 0 ... 0.000000\ntrafficSource.isTrueDirect 11668 ... 68.301821\ntrafficSource.keyword 10542 ... 61.710472\ntrafficSource.medium 0 ... 0.000000\ntrafficSource.referralPath 11460 ... 67.084236\ntrafficSource.source 0 ... 0.000000\nvisitId 0 ... 0.000000\nvisitNumber 0 ... 0.000000\nvisitStartTime 0 ... 0.000000\n\n[59 rows x 3 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
MissingCountPercent
channelGrouping0170830.000000
customDimensions0170830.000000
date0170830.000000
device.browser0170830.000000
device.browserSize0170830.000000
device.browserVersion0170830.000000
device.deviceCategory0170830.000000
device.flashVersion0170830.000000
device.isMobile0170830.000000
device.language0170830.000000
device.mobileDeviceBranding0170830.000000
device.mobileDeviceInfo0170830.000000
device.mobileDeviceMarketingName0170830.000000
device.mobileDeviceModel0170830.000000
device.mobileInputSelector0170830.000000
device.operatingSystem0170830.000000
device.operatingSystemVersion0170830.000000
device.screenColors0170830.000000
device.screenResolution0170830.000000
fullVisitorId0170830.000000
geoNetwork.city0170830.000000
geoNetwork.cityId0170830.000000
geoNetwork.continent0170830.000000
geoNetwork.country0170830.000000
geoNetwork.latitude0170830.000000
geoNetwork.longitude0170830.000000
geoNetwork.metro0170830.000000
geoNetwork.networkDomain0170830.000000
geoNetwork.networkLocation0170830.000000
geoNetwork.region0170830.000000
geoNetwork.subContinent0170830.000000
hits0170830.000000
socialEngagementType0170830.000000
totals.bounces8290879348.527776
totals.hits0170830.000000
totals.newVisits39731311023.257039
totals.pageviews4170790.023415
totals.sessionQualityDim8309877448.638998
totals.timeOnSite8829825451.682960
totals.totalTransactionRevenue1693015399.104373
totals.transactionRevenue1693015399.104373
totals.transactions1692915499.098519
totals.visits0170830.000000
trafficSource.adContent1643764696.218463
trafficSource.adwordsClickInfo.adNetworkType1633375095.609670
trafficSource.adwordsClickInfo.criteriaParameters0170830.000000
trafficSource.adwordsClickInfo.gclId1633275195.603817
trafficSource.adwordsClickInfo.isVideoAd1633375095.609670
trafficSource.adwordsClickInfo.page1633375095.609670
trafficSource.adwordsClickInfo.slot1633375095.609670
trafficSource.campaign0170830.000000
trafficSource.isTrueDirect11668541568.301821
trafficSource.keyword10542654161.710472
trafficSource.medium0170830.000000
trafficSource.referralPath11460562367.084236
trafficSource.source0170830.000000
visitId0170830.000000
visitNumber0170830.000000
visitStartTime0170830.000000
\n
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"missing_values(testingData)","execution_count":49,"outputs":[{"output_type":"stream","text":" Missing ... Percent\ntotals.bounces 2177 ... 54.221669\ntotals.newVisits 1134 ... 28.244085\ntotals.pageviews 2 ... 0.049813\ntotals.timeOnSite 1853 ... 46.151930\ntotals.totalTransactionRevenue 3960 ... 98.630137\ntotals.transactionRevenue 3960 ... 98.630137\ntotals.transactions 3941 ... 98.156912\ntrafficSource.adwordsClickInfo.adNetworkType 3908 ... 97.334994\ntrafficSource.adwordsClickInfo.gclId 3908 ... 97.334994\ntrafficSource.adwordsClickInfo.isVideoAd 3908 ... 97.334994\ntrafficSource.adwordsClickInfo.page 3908 ... 97.334994\ntrafficSource.adwordsClickInfo.slot 3908 ... 97.334994\ntrafficSource.isTrueDirect 2546 ... 63.412204\ntrafficSource.keyword 400 ... 9.962640\n\n[14 rows x 3 columns]\n","name":"stdout"},{"output_type":"stream","text":"/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:10: FutureWarning:\n\nSorting because non-concatenation axis is not aligned. A future version\nof pandas will change to not sort by default.\n\nTo accept the future behavior, pass 'sort=False'.\n\nTo retain the current behavior and silence the warning, pass 'sort=True'.\n\n\n","name":"stderr"},{"output_type":"execute_result","execution_count":49,"data":{"text/plain":" Missing ... Percent\nchannelGrouping 0 ... 0.000000\ncustomDimensions 0 ... 0.000000\ndate 0 ... 0.000000\ndevice.browser 0 ... 0.000000\ndevice.browserSize 0 ... 0.000000\ndevice.browserVersion 0 ... 0.000000\ndevice.deviceCategory 0 ... 0.000000\ndevice.flashVersion 0 ... 0.000000\ndevice.isMobile 0 ... 0.000000\ndevice.language 0 ... 0.000000\ndevice.mobileDeviceBranding 0 ... 0.000000\ndevice.mobileDeviceInfo 0 ... 0.000000\ndevice.mobileDeviceMarketingName 0 ... 0.000000\ndevice.mobileDeviceModel 0 ... 0.000000\ndevice.mobileInputSelector 0 ... 0.000000\ndevice.operatingSystem 0 ... 0.000000\ndevice.operatingSystemVersion 0 ... 0.000000\ndevice.screenColors 0 ... 0.000000\ndevice.screenResolution 0 ... 0.000000\nfullVisitorId 0 ... 0.000000\ngeoNetwork.city 0 ... 0.000000\ngeoNetwork.cityId 0 ... 0.000000\ngeoNetwork.continent 0 ... 0.000000\ngeoNetwork.country 0 ... 0.000000\ngeoNetwork.latitude 0 ... 0.000000\ngeoNetwork.longitude 0 ... 0.000000\ngeoNetwork.metro 0 ... 0.000000\ngeoNetwork.networkDomain 0 ... 0.000000\ngeoNetwork.networkLocation 0 ... 0.000000\ngeoNetwork.region 0 ... 0.000000\ngeoNetwork.subContinent 0 ... 0.000000\nhits 0 ... 0.000000\nsocialEngagementType 0 ... 0.000000\ntotals.bounces 2177 ... 54.221669\ntotals.hits 0 ... 0.000000\ntotals.newVisits 1134 ... 28.244085\ntotals.pageviews 2 ... 0.049813\ntotals.sessionQualityDim 0 ... 0.000000\ntotals.timeOnSite 1853 ... 46.151930\ntotals.totalTransactionRevenue 3960 ... 98.630137\ntotals.transactionRevenue 3960 ... 98.630137\ntotals.transactions 3941 ... 98.156912\ntotals.visits 0 ... 0.000000\ntrafficSource.adContent 0 ... 0.000000\ntrafficSource.adwordsClickInfo.adNetworkType 3908 ... 97.334994\ntrafficSource.adwordsClickInfo.criteriaParameters 0 ... 0.000000\ntrafficSource.adwordsClickInfo.gclId 3908 ... 97.334994\ntrafficSource.adwordsClickInfo.isVideoAd 3908 ... 97.334994\ntrafficSource.adwordsClickInfo.page 3908 ... 97.334994\ntrafficSource.adwordsClickInfo.slot 3908 ... 97.334994\ntrafficSource.campaign 0 ... 0.000000\ntrafficSource.isTrueDirect 2546 ... 63.412204\ntrafficSource.keyword 400 ... 9.962640\ntrafficSource.medium 0 ... 0.000000\ntrafficSource.referralPath 0 ... 0.000000\ntrafficSource.source 0 ... 0.000000\nvisitId 0 ... 0.000000\nvisitNumber 0 ... 0.000000\nvisitStartTime 0 ... 0.000000\n\n[59 rows x 3 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
MissingCountPercent
channelGrouping040150.000000
customDimensions040150.000000
date040150.000000
device.browser040150.000000
device.browserSize040150.000000
device.browserVersion040150.000000
device.deviceCategory040150.000000
device.flashVersion040150.000000
device.isMobile040150.000000
device.language040150.000000
device.mobileDeviceBranding040150.000000
device.mobileDeviceInfo040150.000000
device.mobileDeviceMarketingName040150.000000
device.mobileDeviceModel040150.000000
device.mobileInputSelector040150.000000
device.operatingSystem040150.000000
device.operatingSystemVersion040150.000000
device.screenColors040150.000000
device.screenResolution040150.000000
fullVisitorId040150.000000
geoNetwork.city040150.000000
geoNetwork.cityId040150.000000
geoNetwork.continent040150.000000
geoNetwork.country040150.000000
geoNetwork.latitude040150.000000
geoNetwork.longitude040150.000000
geoNetwork.metro040150.000000
geoNetwork.networkDomain040150.000000
geoNetwork.networkLocation040150.000000
geoNetwork.region040150.000000
geoNetwork.subContinent040150.000000
hits040150.000000
socialEngagementType040150.000000
totals.bounces2177183854.221669
totals.hits040150.000000
totals.newVisits1134288128.244085
totals.pageviews240130.049813
totals.sessionQualityDim040150.000000
totals.timeOnSite1853216246.151930
totals.totalTransactionRevenue39605598.630137
totals.transactionRevenue39605598.630137
totals.transactions39417498.156912
totals.visits040150.000000
trafficSource.adContent040150.000000
trafficSource.adwordsClickInfo.adNetworkType390810797.334994
trafficSource.adwordsClickInfo.criteriaParameters040150.000000
trafficSource.adwordsClickInfo.gclId390810797.334994
trafficSource.adwordsClickInfo.isVideoAd390810797.334994
trafficSource.adwordsClickInfo.page390810797.334994
trafficSource.adwordsClickInfo.slot390810797.334994
trafficSource.campaign040150.000000
trafficSource.isTrueDirect2546146963.412204
trafficSource.keyword40036159.962640
trafficSource.medium040150.000000
trafficSource.referralPath040150.000000
trafficSource.source040150.000000
visitId040150.000000
visitNumber040150.000000
visitStartTime040150.000000
\n
"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"#### Trend Analysis\n\nThe features were plotted to gain a visual understanding of where the trends lied with respect to customer revenue. This provides a basis to understand the features prior to modelling. Each set of visualizations explores a particular set of the features (location, date, device, user experience). The plots show interesting findings for each category of features. \n\nDate - Visits and revenue peak just before the new year\n\nLocation - American customers are the most profitable\n\nDevice - The majority of paying customers access the website using their desktop via Chrome\n\nUser experience - As can be expected, higher session quality and increased page visits correlate to higher revenue per customer\n\nThey were plotted in sets of increasingly detailed columns (i.e. continent, subcontinent, network).\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"datavis = trainingData.copy()\ndatavis.head()","execution_count":50,"outputs":[{"output_type":"execute_result","execution_count":50,"data":{"text/plain":" channelGrouping ... trafficSource.source\n0 Direct ... (direct)\n1 Organic Search ... google\n2 Affiliates ... Partners\n3 Organic Search ... bing\n4 Organic Search ... google\n\n[5 rows x 59 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
channelGroupingcustomDimensionsdatefullVisitorIdhitssocialEngagementTypevisitIdvisitNumbervisitStartTimedevice.browserdevice.browserSizedevice.browserVersiondevice.deviceCategorydevice.flashVersiondevice.isMobiledevice.languagedevice.mobileDeviceBrandingdevice.mobileDeviceInfodevice.mobileDeviceMarketingNamedevice.mobileDeviceModeldevice.mobileInputSelectordevice.operatingSystemdevice.operatingSystemVersiondevice.screenColorsdevice.screenResolutiongeoNetwork.citygeoNetwork.cityIdgeoNetwork.continentgeoNetwork.countrygeoNetwork.latitudegeoNetwork.longitudegeoNetwork.metrogeoNetwork.networkDomaingeoNetwork.networkLocationgeoNetwork.regiongeoNetwork.subContinenttotals.bouncestotals.hitstotals.newVisitstotals.pageviewstotals.sessionQualityDimtotals.timeOnSitetotals.totalTransactionRevenuetotals.transactionRevenuetotals.transactionstotals.visitstrafficSource.adContenttrafficSource.adwordsClickInfo.adNetworkTypetrafficSource.adwordsClickInfo.criteriaParameterstrafficSource.adwordsClickInfo.gclIdtrafficSource.adwordsClickInfo.isVideoAdtrafficSource.adwordsClickInfo.pagetrafficSource.adwordsClickInfo.slottrafficSource.campaigntrafficSource.isTrueDirecttrafficSource.keywordtrafficSource.mediumtrafficSource.referralPathtrafficSource.source
0Direct[{'index': '4', 'value': 'North America'}]201710164447569759370230345[{'hitNumber': '1', 'time': '0', 'hour': '18',...Not Socially Engaged150820327911508203279Chromenot available in demo datasetnot available in demo datasetmobilenot available in demo datasetTruenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetAndroidnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetnot available in demo datasetpsu.edunot available in demo datasetnot available in demo datasetNorthern AmericaNaN414166NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaN(not set)TrueNaN(none)NaN(direct)
1Organic Search[{'index': '4', 'value': 'North America'}]201710168449727979333841797[{'hitNumber': '1', 'time': '0', 'hour': '15',...Not Socially Engaged150819259711508192597Chromenot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetLinuxnot available in demo datasetnot available in demo datasetnot available in demo datasetSunnyvalenot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetSan Francisco-Oakland-San Jose CA(not set)not available in demo datasetCaliforniaNorthern AmericaNaN616240NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaN(not set)NaN(not provided)organicNaNgoogle
2Affiliates[{'index': '4', 'value': 'EMEA'}]201710160258560966793548223[{'hitNumber': '1', 'time': '0', 'hour': '7', ...Not Socially Engaged150816375511508163755Internet Explorernot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetWindowsnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetEuropeNetherlandsnot available in demo datasetnot available in demo datasetnot available in demo datasetkpn.netnot available in demo datasetnot available in demo datasetWestern EuropeNaN7171243NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaNData Share PromoNaNNaNaffiliateNaNPartners
3Organic Search[{'index': '4', 'value': 'Central America'}]201710162120129608233101123[{'hitNumber': '1', 'time': '0', 'hour': '11',...Not Socially Engaged150817726111508177261Internet Explorernot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetWindowsnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetAmericasMexiconot available in demo datasetnot available in demo datasetnot available in demo datasetunknown.unknownnot available in demo datasetnot available in demo datasetCentral AmericaNaN8181111NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaN(not set)NaNgoogle online merchandiseorganicNaNbing
4Organic Search[{'index': '4', 'value': 'North America'}]201710163848523864758719206[{'hitNumber': '1', 'time': '0', 'hour': '14',...Not Socially Engaged150818899511508188995Chromenot available in demo datasetnot available in demo datasetdesktopnot available in demo datasetFalsenot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetChrome OSnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetnot available in demo datasetAmericasUnited Statesnot available in demo datasetnot available in demo datasetnot available in demo dataset(not set)not available in demo datasetnot available in demo datasetNorthern AmericaNaN9176250NaNNaNNaN1NaNNaNnot available in demo datasetNaNNaNNaNNaN(not set)NaN(not provided)organicNaNgoogle
\n
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"datavis.info()","execution_count":51,"outputs":[{"output_type":"stream","text":"\nRangeIndex: 17083 entries, 0 to 17082\nData columns (total 59 columns):\nchannelGrouping 17083 non-null object\ncustomDimensions 17083 non-null object\ndate 17083 non-null int64\nfullVisitorId 17083 non-null object\nhits 17083 non-null object\nsocialEngagementType 17083 non-null object\nvisitId 17083 non-null int64\nvisitNumber 17083 non-null int64\nvisitStartTime 17083 non-null int64\ndevice.browser 17083 non-null object\ndevice.browserSize 17083 non-null object\ndevice.browserVersion 17083 non-null object\ndevice.deviceCategory 17083 non-null object\ndevice.flashVersion 17083 non-null object\ndevice.isMobile 17083 non-null bool\ndevice.language 17083 non-null object\ndevice.mobileDeviceBranding 17083 non-null object\ndevice.mobileDeviceInfo 17083 non-null object\ndevice.mobileDeviceMarketingName 17083 non-null object\ndevice.mobileDeviceModel 17083 non-null object\ndevice.mobileInputSelector 17083 non-null object\ndevice.operatingSystem 17083 non-null object\ndevice.operatingSystemVersion 17083 non-null object\ndevice.screenColors 17083 non-null object\ndevice.screenResolution 17083 non-null object\ngeoNetwork.city 17083 non-null object\ngeoNetwork.cityId 17083 non-null object\ngeoNetwork.continent 17083 non-null object\ngeoNetwork.country 17083 non-null object\ngeoNetwork.latitude 17083 non-null object\ngeoNetwork.longitude 17083 non-null object\ngeoNetwork.metro 17083 non-null object\ngeoNetwork.networkDomain 17083 non-null object\ngeoNetwork.networkLocation 17083 non-null object\ngeoNetwork.region 17083 non-null object\ngeoNetwork.subContinent 17083 non-null object\ntotals.bounces 8793 non-null object\ntotals.hits 17083 non-null object\ntotals.newVisits 13110 non-null object\ntotals.pageviews 17079 non-null object\ntotals.sessionQualityDim 8774 non-null object\ntotals.timeOnSite 8254 non-null object\ntotals.totalTransactionRevenue 153 non-null object\ntotals.transactionRevenue 153 non-null object\ntotals.transactions 154 non-null object\ntotals.visits 17083 non-null object\ntrafficSource.adContent 646 non-null object\ntrafficSource.adwordsClickInfo.adNetworkType 750 non-null object\ntrafficSource.adwordsClickInfo.criteriaParameters 17083 non-null object\ntrafficSource.adwordsClickInfo.gclId 751 non-null object\ntrafficSource.adwordsClickInfo.isVideoAd 750 non-null object\ntrafficSource.adwordsClickInfo.page 750 non-null object\ntrafficSource.adwordsClickInfo.slot 750 non-null object\ntrafficSource.campaign 17083 non-null object\ntrafficSource.isTrueDirect 5415 non-null object\ntrafficSource.keyword 6541 non-null object\ntrafficSource.medium 17083 non-null object\ntrafficSource.referralPath 5623 non-null object\ntrafficSource.source 17083 non-null object\ndtypes: bool(1), int64(4), object(54)\nmemory usage: 7.6+ MB\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"datavis[\"totals.transactionRevenue\"] = datavis[\"totals.transactionRevenue\"].astype('float')\ngdf = datavis.groupby(\"fullVisitorId\")[\"totals.transactionRevenue\"].sum().reset_index()\n\nplt.figure(figsize=(8,6))\nplt.scatter(range(gdf.shape[0]), np.sort(np.log1p(gdf[\"totals.transactionRevenue\"].values)))\nplt.xlabel('index', fontsize=12)\nplt.ylabel('TransactionRevenue', fontsize=12)\nplt.show()","execution_count":52,"outputs":[{"output_type":"display_data","data":{"text/plain":"
","image/png":"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\n"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"nzi = pd.notnull(datavis[\"totals.transactionRevenue\"]).sum()\npaying = (gdf[\"totals.transactionRevenue\"]>0).sum()\nprint(\"There are\",paying,\"paying customers\")\nimport plotly.plotly as py\nimport plotly.graph_objs as go\nfrom plotly import tools\nprint(\"Percentage of customers that generate revenue is\",round(paying*100 / gdf.shape[0],2),\"%\")","execution_count":53,"outputs":[{"output_type":"stream","text":"There are 152 paying customers\nPercentage of customers that generate revenue is 0.9 %\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"def scatter_plot(cnt_srs, color):\n trace = go.Scatter(\n x=cnt_srs.index[::-1],\n y=cnt_srs.values[::-1],\n showlegend=False,\n marker=dict(\n color=color,\n ),\n )\n return trace\ndatavis['date'] = datavis['date'].apply(lambda x: dt.date(int(str(x)[:4]), int(str(x)[4:6]), int(str(x)[6:])))\ncnt_srs = datavis.groupby('date')['totals.transactionRevenue'].agg(['size', 'count'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\"]\ncnt_srs = cnt_srs.sort_index()\ncnt_srs.index = cnt_srs.index.astype('str')\ntrace1 = scatter_plot(cnt_srs[\"count\"], 'red')\ntrace2 = scatter_plot(cnt_srs[\"count of non-zero revenue\"], 'blue')\n\nfig = tools.make_subplots(rows=2, cols=1, vertical_spacing=0.08,\n subplot_titles=[\"Date - Visitors\", \"Date - Paying Customers\"])\nfig.append_trace(trace1, 1, 1)\nfig.append_trace(trace2, 2, 1)\nfig['layout'].update(height=800, width=800, paper_bgcolor='rgb(233,233,233)', title=\"Date Plots\")\ninit_notebook_mode(connected=True)\niplot(fig, filename='date-plots')","execution_count":54,"outputs":[{"output_type":"stream","text":"This is the format of your plot grid:\n[ (1,1) x1,y1 ]\n[ (2,1) x2,y2 ]\n\n","name":"stdout"},{"output_type":"display_data","data":{"text/html":"","text/vnd.plotly.v1+html":""},"metadata":{}},{"output_type":"display_data","data":{"application/vnd.plotly.v1+json":{"data":[{"marker":{"color":"red"},"showlegend":false,"x":["2018-04-30","2018-04-29","2018-04-28","2018-04-27","2018-04-26","2018-04-25","2018-04-24","2018-04-23","2018-04-22","2018-04-21","2018-04-20","2018-04-19","2018-04-18","2018-04-17","2018-04-16","2018-04-15","2018-04-14","2018-04-13","2018-04-12","2018-04-11","2018-04-10","2018-04-09","2018-04-08","2018-04-07","2018-04-06","2018-04-05","2018-04-04","2018-04-03","2018-04-02","2018-04-01","2018-03-31","2018-03-30","2018-03-29","2018-03-28","2018-03-27","2018-03-26","2018-03-25","2018-03-24","2018-03-23","2018-03-22","2018-03-21","2018-03-20","2018-03-19","2018-03-18","2018-03-17","2018-03-16","2018-03-15","2018-03-14","2018-03-13","2018-03-12","2018-03-11","2018-03-10","2018-03-09","2018-03-08","2018-03-07","2018-03-06","2018-03-05","2018-03-04","2018-03-03","2018-03-02","2018-03-01","2018-02-28","2018-02-27","2018-02-26","2018-02-25","2018-02-24","2018-02-23","2018-02-22","2018-02-21","2018-02-20","2018-02-19","2018-02-18","2018-02-17","2018-02-16","2018-02-15","2018-02-14","2018-02-13","2018-02-12","2018-02-11","2018-02-10","2018-02-09","2018-02-08","2018-02-07","2018-02-06","2018-02-05","2018-02-04","2018-02-03","2018-02-02","2018-02-01","2018-01-31","2018-01-30","2018-01-29","2018-01-28","2018-01-27","2018-01-26","2018-01-25","2018-01-24","2018-01-23","2018-01-22","2018-01-21","2018-01-20","2018-01-19","2018-01-18","2018-01-17","2018-01-16","2018-01-15","2018-01-14","2018-01-13","2018-01-12","2018-01-11","2018-01-10","2018-01-09","2018-01-08","2018-01-07","2018-01-06","2018-01-05","2018-01-04","2018-01-03","2018-01-02","2018-01-01","2017-12-31","2017-12-30","2017-12-29","2017-12-28","2017-12-27","2017-12-26","2017-12-25","2017-12-24","2017-12-23","2017-12-22","2017-12-21","2017-12-20","2017-12-19","2017-12-18","2017-12-17","2017-12-16","2017-12-15","2017-12-14","2017-12-13","2017-12-12","2017-12-11","2017-12-10","2017-12-09","2017-12-08","2017-12-07","2017-12-06","2017-12-05","2017-12-04","2017-12-03","2017-12-02","2017-12-01","2017-11-30","2017-11-29","2017-11-28","2017-11-27","2017-11-26","2017-11-25","2017-11-24","2017-11-23","2017-11-22","2017-11-21","2017-11-20","2017-11-19","2017-11-18","2017-11-17","2017-11-16","2017-11-15","2017-11-14","2017-11-13","2017-11-12","2017-11-11","2017-11-10","2017-11-09","2017-11-08","2017-11-07","2017-11-06","2017-11-05","2017-11-04","2017-11-03","2017-11-02","2017-11-01","2017-10-31","2017-10-30","2017-10-29","2017-10-28","2017-10-27","2017-10-26","2017-10-25","2017-10-24","2017-10-23","2017-10-22","2017-10-21","2017-10-20","2017-10-19","2017-10-18","2017-10-17","2017-10-16","2017-10-15","2017-10-14","2017-10-13","2017-10-12","2017-10-11","2017-10-10","2017-10-09","2017-10-08","2017-10-07","2017-10-06","2017-10-05","2017-10-04","2017-10-03","2017-10-02","2017-10-01","2017-09-30","2017-09-29","2017-09-28","2017-09-27","2017-09-26","2017-09-25","2017-09-24","2017-09-23","2017-09-22","2017-09-21","2017-09-20","2017-09-19","2017-09-18","2017-09-17","2017-09-16","2017-09-15","2017-09-14","2017-09-13","2017-09-12","2017-09-11","2017-09-10","2017-09-09","2017-09-08","2017-09-07","2017-09-06","2017-09-05","2017-09-04","2017-09-03","2017-09-02","2017-09-01","2017-08-31","2017-08-30","2017-08-29","2017-08-28","2017-08-27","2017-08-26","2017-08-25","2017-08-24","2017-08-23","2017-08-22","2017-08-21","2017-08-20","2017-08-19","2017-08-18","2017-08-17","2017-08-16","2017-08-15","2017-08-14","2017-08-13","2017-08-12","2017-08-11","2017-08-10","2017-08-09","2017-08-08","2017-08-07","2017-08-06","2017-08-05","2017-08-04","2017-08-03","2017-08-02","2017-08-01","2017-07-31","2017-07-30","2017-07-29","2017-07-28","2017-07-27","2017-07-26","2017-07-25","2017-07-24","2017-07-23","2017-07-22","2017-07-21","2017-07-20","2017-07-19","2017-07-18","2017-07-17","2017-07-16","2017-07-15","2017-07-14","2017-07-13","2017-07-12","2017-07-11","2017-07-10","2017-07-09","2017-07-08","2017-07-07","2017-07-06","2017-07-05","2017-07-04","2017-07-03","2017-07-02","2017-07-01","2017-06-30","2017-06-29","2017-06-28","2017-06-27","2017-06-26","2017-06-25","2017-06-24","2017-06-23","2017-06-22","2017-06-21","2017-06-20","2017-06-19","2017-06-18","2017-06-17","2017-06-16","2017-06-15","2017-06-14","2017-06-13","2017-06-12","2017-06-11","2017-06-10","2017-06-09","2017-06-08","2017-06-07","2017-06-06","2017-06-05","2017-06-04","2017-06-03","2017-06-02","2017-06-01","2017-05-31","2017-05-30","2017-05-29","2017-05-28","2017-05-27","2017-05-26","2017-05-25","2017-05-24","2017-05-23","2017-05-22","2017-05-21","2017-05-20","2017-05-19","2017-05-18","2017-05-17","2017-05-16","2017-05-15","2017-05-14","2017-05-13","2017-05-12","2017-05-11","2017-05-10","2017-05-09","2017-05-08","2017-05-07","2017-05-06","2017-05-05","2017-05-04","2017-05-03","2017-05-02","2017-05-01","2017-04-30","2017-04-29","2017-04-28","2017-04-27","2017-04-26","2017-04-25","2017-04-24","2017-04-23","2017-04-22","2017-04-21","2017-04-20","2017-04-19","2017-04-18","2017-04-17","2017-04-16","2017-04-15","2017-04-14","2017-04-13","2017-04-12","2017-04-11","2017-04-10","2017-04-09","2017-04-08","2017-04-07","2017-04-06","2017-04-05","2017-04-04","2017-04-03","2017-04-02","2017-04-01","2017-03-31","2017-03-30","2017-03-29","2017-03-28","2017-03-27","2017-03-26","2017-03-25","2017-03-24","2017-03-23","2017-03-22","2017-03-21","2017-03-20","2017-03-19","2017-03-18","2017-03-17","2017-03-16","2017-03-15","2017-03-14","2017-03-13","2017-03-12","2017-03-11","2017-03-10","2017-03-09","2017-03-08","2017-03-07","2017-03-06","2017-03-05","2017-03-04","2017-03-03","2017-03-02","2017-03-01","2017-02-28","2017-02-27","2017-02-26","2017-02-25","2017-02-24","2017-02-23","2017-02-22","2017-02-21","2017-02-20","2017-02-19","2017-02-18","2017-02-17","2017-02-16","2017-02-15","2017-02-14","2017-02-13","2017-02-12","2017-02-11","2017-02-10","2017-02-09","2017-02-08","2017-02-07","2017-02-06","2017-02-05","2017-02-04","2017-02-03","2017-02-02","2017-02-01","2017-01-31","2017-01-30","2017-01-29","2017-01-28","2017-01-27","2017-01-26","2017-01-25","2017-01-24","2017-01-23","2017-01-22","2017-01-21","2017-01-20","2017-01-19","2017-01-18","2017-01-17","2017-01-16","2017-01-15","2017-01-14","2017-01-13","2017-01-12","2017-01-11","2017-01-10","2017-01-09","2017-01-08","2017-01-07","2017-01-06","2017-01-05","2017-01-04","2017-01-03","2017-01-02","2017-01-01","2016-12-31","2016-12-30","2016-12-29","2016-12-28","2016-12-27","2016-12-26","2016-12-25","2016-12-24","2016-12-23","2016-12-22","2016-12-21","2016-12-20","2016-12-19","2016-12-18","2016-12-17","2016-12-16","2016-12-15","2016-12-14","2016-12-13","2016-12-12","2016-12-11","2016-12-10","2016-12-09","2016-12-08","2016-12-07","2016-12-06","2016-12-05","2016-12-04","2016-12-03","2016-12-02","2016-12-01","2016-11-30","2016-11-29","2016-11-28","2016-11-27","2016-11-26","2016-11-25","2016-11-24","2016-11-23","2016-11-22","2016-11-21","2016-11-20","2016-11-19","2016-11-18","2016-11-17","2016-11-16","2016-11-15","2016-11-14","2016-11-13","2016-11-12","2016-11-11","2016-11-10","2016-11-09","2016-11-08","2016-11-07","2016-11-06","2016-11-05","2016-11-04","2016-11-03","2016-11-02","2016-11-01","2016-10-31","2016-10-30","2016-10-29","2016-10-28","2016-10-27","2016-10-26","2016-10-25","2016-10-24","2016-10-23","2016-10-22","2016-10-21","2016-10-20","2016-10-19","2016-10-18","2016-10-17","2016-10-16","2016-10-15","2016-10-14","2016-10-13","2016-10-12","2016-10-11","2016-10-10","2016-10-09","2016-10-08","2016-10-07","2016-10-06","2016-10-05","2016-10-04","2016-10-03","2016-10-02","2016-10-01","2016-09-30","2016-09-29","2016-09-28","2016-09-27","2016-09-26","2016-09-25","2016-09-24","2016-09-23","2016-09-22","2016-09-21","2016-09-20","2016-09-19","2016-09-18","2016-09-17","2016-09-16","2016-09-15","2016-09-14","2016-09-13","2016-09-12","2016-09-11","2016-09-10","2016-09-09","2016-09-08","2016-09-07","2016-09-06","2016-09-05","2016-09-04","2016-09-03","2016-09-02","2016-09-01","2016-08-31","2016-08-30","2016-08-29","2016-08-28","2016-08-27","2016-08-26","2016-08-25","2016-08-24","2016-08-23","2016-08-22","2016-08-21","2016-08-20","2016-08-19","2016-08-18","2016-08-17","2016-08-16","2016-08-15","2016-08-14","2016-08-13","2016-08-12","2016-08-11","2016-08-10","2016-08-09","2016-08-08","2016-08-07","2016-08-06","2016-08-05","2016-08-04","2016-08-03","2016-08-02","2016-08-01"],"y":[36,19,27,24,29,33,33,37,22,16,24,38,27,24,35,20,15,31,38,39,26,39,28,27,27,36,35,30,28,23,14,19,34,42,37,34,15,18,32,37,31,23,36,13,21,27,47,24,39,34,19,30,34,23,46,51,35,21,22,32,30,38,35,40,23,25,41,24,43,36,30,32,27,22,39,33,43,34,36,22,29,28,23,26,33,33,21,25,36,31,36,32,24,22,29,31,23,30,27,14,17,31,46,29,36,33,22,17,40,34,32,28,19,23,19,24,29,27,31,12,11,19,19,29,33,17,28,29,21,22,36,33,34,28,25,30,32,35,94,77,34,18,24,38,31,32,43,31,24,19,32,27,30,41,36,18,21,29,29,37,28,28,25,22,27,46,38,44,36,25,18,35,35,36,48,37,26,23,27,29,28,26,25,16,21,23,31,45,33,36,25,23,23,23,30,46,24,18,17,29,38,53,52,30,30,36,48,49,42,35,39,30,29,24,41,30,36,32,36,23,37,50,47,46,35,37,28,20,32,32,26,31,22,17,31,30,26,22,27,16,25,19,31,35,35,20,24,27,30,34,34,25,27,15,15,25,26,29,28,35,18,25,22,24,32,29,36,25,17,27,32,26,21,28,12,10,25,19,23,25,20,13,13,24,27,31,29,21,20,15,31,24,18,22,32,21,17,30,26,25,23,30,14,18,27,21,30,30,18,14,21,17,27,23,28,26,13,11,12,21,26,21,30,13,14,15,24,33,20,33,13,12,20,35,32,23,16,8,16,20,24,23,22,24,23,14,31,28,44,32,29,4,19,20,23,8,13,19,12,12,19,27,16,33,27,17,15,30,37,40,20,21,18,15,22,29,21,24,25,14,18,25,20,18,27,24,16,21,19,24,24,25,15,17,14,23,29,22,29,33,18,16,24,28,24,30,28,22,17,27,30,24,19,26,19,17,22,21,28,22,18,20,19,20,32,28,15,17,21,16,17,16,26,24,25,24,15,24,27,22,30,23,18,14,21,24,24,18,21,12,20,22,16,24,24,21,14,21,23,33,35,25,16,17,10,16,27,15,30,17,13,14,27,23,25,38,26,17,14,26,28,20,16,10,12,17,13,21,24,23,13,20,14,13,28,29,29,47,22,25,28,37,31,35,31,24,22,32,28,42,19,43,28,29,33,53,45,50,50,38,17,35,34,39,41,42,35,30,37,33,35,51,35,32,34,33,48,39,38,38,37,27,36,42,55,34,26,37,21,39,34,48,37,32,25,28,34,33,37,39,38,21,35,32,22,35,34,30,18,14,35,30,40,42,18,18,15,17,24,26,18,17,17,17,24,32,21,32,23,21,17,22,25,22,19,21,24,14,19,21,23,20,17,10,14,22,25,30,31,26,18,19,29,20,18,27,34,22,14,30,31,27,32,29,16,15,17,23,31,28,38,12,14,22,23,39,28,12],"type":"scatter","uid":"6d461e0b-4339-4594-b59e-34ab661ea7a0","xaxis":"x","yaxis":"y"},{"marker":{"color":"blue"},"showlegend":false,"x":["2018-04-30","2018-04-29","2018-04-28","2018-04-27","2018-04-26","2018-04-25","2018-04-24","2018-04-23","2018-04-22","2018-04-21","2018-04-20","2018-04-19","2018-04-18","2018-04-17","2018-04-16","2018-04-15","2018-04-14","2018-04-13","2018-04-12","2018-04-11","2018-04-10","2018-04-09","2018-04-08","2018-04-07","2018-04-06","2018-04-05","2018-04-04","2018-04-03","2018-04-02","2018-04-01","2018-03-31","2018-03-30","2018-03-29","2018-03-28","2018-03-27","2018-03-26","2018-03-25","2018-03-24","2018-03-23","2018-03-22","2018-03-21","2018-03-20","2018-03-19","2018-03-18","2018-03-17","2018-03-16","2018-03-15","2018-03-14","2018-03-13","2018-03-12","2018-03-11","2018-03-10","2018-03-09","2018-03-08","2018-03-07","2018-03-06","2018-03-05","2018-03-04","2018-03-03","2018-03-02","2018-03-01","2018-02-28","2018-02-27","2018-02-26","2018-02-25","2018-02-24","2018-02-23","2018-02-22","2018-02-21","2018-02-20","2018-02-19","2018-02-18","2018-02-17","2018-02-16","2018-02-15","2018-02-14","2018-02-13","2018-02-12","2018-02-11","2018-02-10","2018-02-09","2018-02-08","2018-02-07","2018-02-06","2018-02-05","2018-02-04","2018-02-03","2018-02-02","2018-02-01","2018-01-31","2018-01-30","2018-01-29","2018-01-28","2018-01-27","2018-01-26","2018-01-25","2018-01-24","2018-01-23","2018-01-22","2018-01-21","2018-01-20","2018-01-19","2018-01-18","2018-01-17","2018-01-16","2018-01-15","2018-01-14","2018-01-13","2018-01-12","2018-01-11","2018-01-10","2018-01-09","2018-01-08","2018-01-07","2018-01-06","2018-01-05","2018-01-04","2018-01-03","2018-01-02","2018-01-01","2017-12-31","2017-12-30","2017-12-29","2017-12-28","2017-12-27","2017-12-26","2017-12-25","2017-12-24","2017-12-23","2017-12-22","2017-12-21","2017-12-20","2017-12-19","2017-12-18","2017-12-17","2017-12-16","2017-12-15","2017-12-14","2017-12-13","2017-12-12","2017-12-11","2017-12-10","2017-12-09","2017-12-08","2017-12-07","2017-12-06","2017-12-05","2017-12-04","2017-12-03","2017-12-02","2017-12-01","2017-11-30","2017-11-29","2017-11-28","2017-11-27","2017-11-26","2017-11-25","2017-11-24","2017-11-23","2017-11-22","2017-11-21","2017-11-20","2017-11-19","2017-11-18","2017-11-17","2017-11-16","2017-11-15","2017-11-14","2017-11-13","2017-11-12","2017-11-11","2017-11-10","2017-11-09","2017-11-08","2017-11-07","2017-11-06","2017-11-05","2017-11-04","2017-11-03","2017-11-02","2017-11-01","2017-10-31","2017-10-30","2017-10-29","2017-10-28","2017-10-27","2017-10-26","2017-10-25","2017-10-24","2017-10-23","2017-10-22","2017-10-21","2017-10-20","2017-10-19","2017-10-18","2017-10-17","2017-10-16","2017-10-15","2017-10-14","2017-10-13","2017-10-12","2017-10-11","2017-10-10","2017-10-09","2017-10-08","2017-10-07","2017-10-06","2017-10-05","2017-10-04","2017-10-03","2017-10-02","2017-10-01","2017-09-30","2017-09-29","2017-09-28","2017-09-27","2017-09-26","2017-09-25","2017-09-24","2017-09-23","2017-09-22","2017-09-21","2017-09-20","2017-09-19","2017-09-18","2017-09-17","2017-09-16","2017-09-15","2017-09-14","2017-09-13","2017-09-12","2017-09-11","2017-09-10","2017-09-09","2017-09-08","2017-09-07","2017-09-06","2017-09-05","2017-09-04","2017-09-03","2017-09-02","2017-09-01","2017-08-31","2017-08-30","2017-08-29","2017-08-28","2017-08-27","2017-08-26","2017-08-25","2017-08-24","2017-08-23","2017-08-22","2017-08-21","2017-08-20","2017-08-19","2017-08-18","2017-08-17","2017-08-16","2017-08-15","2017-08-14","2017-08-13","2017-08-12","2017-08-11","2017-08-10","2017-08-09","2017-08-08","2017-08-07","2017-08-06","2017-08-05","2017-08-04","2017-08-03","2017-08-02","2017-08-01","2017-07-31","2017-07-30","2017-07-29","2017-07-28","2017-07-27","2017-07-26","2017-07-25","2017-07-24","2017-07-23","2017-07-22","2017-07-21","2017-07-20","2017-07-19","2017-07-18","2017-07-17","2017-07-16","2017-07-15","2017-07-14","2017-07-13","2017-07-12","2017-07-11","2017-07-10","2017-07-09","2017-07-08","2017-07-07","2017-07-06","2017-07-05","2017-07-04","2017-07-03","2017-07-02","2017-07-01","2017-06-30","2017-06-29","2017-06-28","2017-06-27","2017-06-26","2017-06-25","2017-06-24","2017-06-23","2017-06-22","2017-06-21","2017-06-20","2017-06-19","2017-06-18","2017-06-17","2017-06-16","2017-06-15","2017-06-14","2017-06-13","2017-06-12","2017-06-11","2017-06-10","2017-06-09","2017-06-08","2017-06-07","2017-06-06","2017-06-05","2017-06-04","2017-06-03","2017-06-02","2017-06-01","2017-05-31","2017-05-30","2017-05-29","2017-05-28","2017-05-27","2017-05-26","2017-05-25","2017-05-24","2017-05-23","2017-05-22","2017-05-21","2017-05-20","2017-05-19","2017-05-18","2017-05-17","2017-05-16","2017-05-15","2017-05-14","2017-05-13","2017-05-12","2017-05-11","2017-05-10","2017-05-09","2017-05-08","2017-05-07","2017-05-06","2017-05-05","2017-05-04","2017-05-03","2017-05-02","2017-05-01","2017-04-30","2017-04-29","2017-04-28","2017-04-27","2017-04-26","2017-04-25","2017-04-24","2017-04-23","2017-04-22","2017-04-21","2017-04-20","2017-04-19","2017-04-18","2017-04-17","2017-04-16","2017-04-15","2017-04-14","2017-04-13","2017-04-12","2017-04-11","2017-04-10","2017-04-09","2017-04-08","2017-04-07","2017-04-06","2017-04-05","2017-04-04","2017-04-03","2017-04-02","2017-04-01","2017-03-31","2017-03-30","2017-03-29","2017-03-28","2017-03-27","2017-03-26","2017-03-25","2017-03-24","2017-03-23","2017-03-22","2017-03-21","2017-03-20","2017-03-19","2017-03-18","2017-03-17","2017-03-16","2017-03-15","2017-03-14","2017-03-13","2017-03-12","2017-03-11","2017-03-10","2017-03-09","2017-03-08","2017-03-07","2017-03-06","2017-03-05","2017-03-04","2017-03-03","2017-03-02","2017-03-01","2017-02-28","2017-02-27","2017-02-26","2017-02-25","2017-02-24","2017-02-23","2017-02-22","2017-02-21","2017-02-20","2017-02-19","2017-02-18","2017-02-17","2017-02-16","2017-02-15","2017-02-14","2017-02-13","2017-02-12","2017-02-11","2017-02-10","2017-02-09","2017-02-08","2017-02-07","2017-02-06","2017-02-05","2017-02-04","2017-02-03","2017-02-02","2017-02-01","2017-01-31","2017-01-30","2017-01-29","2017-01-28","2017-01-27","2017-01-26","2017-01-25","2017-01-24","2017-01-23","2017-01-22","2017-01-21","2017-01-20","2017-01-19","2017-01-18","2017-01-17","2017-01-16","2017-01-15","2017-01-14","2017-01-13","2017-01-12","2017-01-11","2017-01-10","2017-01-09","2017-01-08","2017-01-07","2017-01-06","2017-01-05","2017-01-04","2017-01-03","2017-01-02","2017-01-01","2016-12-31","2016-12-30","2016-12-29","2016-12-28","2016-12-27","2016-12-26","2016-12-25","2016-12-24","2016-12-23","2016-12-22","2016-12-21","2016-12-20","2016-12-19","2016-12-18","2016-12-17","2016-12-16","2016-12-15","2016-12-14","2016-12-13","2016-12-12","2016-12-11","2016-12-10","2016-12-09","2016-12-08","2016-12-07","2016-12-06","2016-12-05","2016-12-04","2016-12-03","2016-12-02","2016-12-01","2016-11-30","2016-11-29","2016-11-28","2016-11-27","2016-11-26","2016-11-25","2016-11-24","2016-11-23","2016-11-22","2016-11-21","2016-11-20","2016-11-19","2016-11-18","2016-11-17","2016-11-16","2016-11-15","2016-11-14","2016-11-13","2016-11-12","2016-11-11","2016-11-10","2016-11-09","2016-11-08","2016-11-07","2016-11-06","2016-11-05","2016-11-04","2016-11-03","2016-11-02","2016-11-01","2016-10-31","2016-10-30","2016-10-29","2016-10-28","2016-10-27","2016-10-26","2016-10-25","2016-10-24","2016-10-23","2016-10-22","2016-10-21","2016-10-20","2016-10-19","2016-10-18","2016-10-17","2016-10-16","2016-10-15","2016-10-14","2016-10-13","2016-10-12","2016-10-11","2016-10-10","2016-10-09","2016-10-08","2016-10-07","2016-10-06","2016-10-05","2016-10-04","2016-10-03","2016-10-02","2016-10-01","2016-09-30","2016-09-29","2016-09-28","2016-09-27","2016-09-26","2016-09-25","2016-09-24","2016-09-23","2016-09-22","2016-09-21","2016-09-20","2016-09-19","2016-09-18","2016-09-17","2016-09-16","2016-09-15","2016-09-14","2016-09-13","2016-09-12","2016-09-11","2016-09-10","2016-09-09","2016-09-08","2016-09-07","2016-09-06","2016-09-05","2016-09-04","2016-09-03","2016-09-02","2016-09-01","2016-08-31","2016-08-30","2016-08-29","2016-08-28","2016-08-27","2016-08-26","2016-08-25","2016-08-24","2016-08-23","2016-08-22","2016-08-21","2016-08-20","2016-08-19","2016-08-18","2016-08-17","2016-08-16","2016-08-15","2016-08-14","2016-08-13","2016-08-12","2016-08-11","2016-08-10","2016-08-09","2016-08-08","2016-08-07","2016-08-06","2016-08-05","2016-08-04","2016-08-03","2016-08-02","2016-08-01"],"y":[1,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,2,2,3,0,0,0,0,0,1,0,0,0,1,0,1,0,0,1,0,0,1,0,1,1,0,0,0,0,0,0,0,0,0,0,0,1,0,0,1,1,0,0,1,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,1,0,0,0,1,0,0,0,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,1,0,0,0,2,0,0,0,0,0,0,0,1,0,1,0,0,0,0,0,2,0,0,1,0,2,0,0,0,0,0,0,0,1,0,0,0,1,0,0,2,0,0,0,0,0,0,0,1,0,1,0,0,0,1,0,0,1,1,1,0,0,0,0,2,0,0,0,1,0,0,1,0,0,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,1,1,0,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,1,0,2,0,0,0,0,2,0,1,0,0,0,2,1,0,0,0,1,0,0,1,0,1,1,0,0,1,0,0,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,3,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,1,0,1,0,0,0,0,1,1,0,0,0,0,1,0,0,0,0,0,0,2,1,1,0,0,0,1,0,0,0,0,1,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,1,0,0,0,0,0,0,0,2,0,0,0,1,0,0,1,0,0,0,0,0,0,0,1,1,1,0,1,1,0,1,0,1,0,0,0,0,1,2,2,0,0,1,0,0,0,0,1,0,0,0,0,1,0,1,0,1,1,1,1,1,1,0,0,0,0,0,0,0,0,1,0,2,0,0,0,0,0,0,0,1,0,0,0,0,1,0,0,1,0,0,0,0,0,0,0,0,0,0,3,2,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0,0,1,0,1,1,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,1,0],"type":"scatter","uid":"d68d869e-914c-4310-a411-db2fe7b82288","xaxis":"x2","yaxis":"y2"}],"layout":{"annotations":[{"font":{"size":16},"showarrow":false,"text":"Date - Visitors","x":0.5,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Date - Paying Customers","x":0.5,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"}],"xaxis":{"anchor":"y","domain":[0,1]},"yaxis":{"anchor":"x","domain":[0.54,1]},"xaxis2":{"anchor":"y2","domain":[0,1]},"yaxis2":{"anchor":"x2","domain":[0,0.46]},"height":800,"width":800,"paper_bgcolor":"rgb(233,233,233)","title":{"text":"Date Plots"}},"config":{"showLink":false,"linkText":"Export to plot.ly","plotlyServerURL":"https://plot.ly"}},"text/html":"
","text/vnd.plotly.v1+html":"
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"def horizontal_bar_chart(cnt_srs, color):\n trace = go.Bar(\n y=cnt_srs.index[::-1],\n x=cnt_srs.values[::-1],\n showlegend=False,\n orientation = 'h',\n marker=dict(\n color=color,\n ),\n )\n return trace\n\n# Device Browser\ncnt_srs = datavis.groupby('device.browser')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace1 = horizontal_bar_chart(cnt_srs[\"count\"].head(10), 'rgba(149, 48, 90, 0.6)')\ntrace2 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(10), 'rgba(149, 48, 90, 0.6)')\ntrace3 = horizontal_bar_chart(cnt_srs[\"mean\"].head(10), 'rgba(149, 48, 90, 0.6)')\n\n# Device Category\ncnt_srs = datavis.groupby('device.deviceCategory')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace4 = horizontal_bar_chart(cnt_srs[\"count\"].head(10), 'rgba(28, 151, 89, 0.6)')\ntrace5 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(10), 'rgba(28, 151, 89, 0.6)')\ntrace6 = horizontal_bar_chart(cnt_srs[\"mean\"].head(10), 'rgba(28, 151, 89, 0.6)')\n\n# Operating system\ncnt_srs = datavis.groupby('device.operatingSystem')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace7 = horizontal_bar_chart(cnt_srs[\"count\"].head(10), 'rgba(60, 150, 60, 0.6)')\ntrace8 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(10),'rgba(60, 150, 60, 0.6)')\ntrace9 = horizontal_bar_chart(cnt_srs[\"mean\"].head(10),'rgba(60, 150, 60, 0.6)')\n\n# Creating two subplots\nfig = tools.make_subplots(rows=3, cols=3, vertical_spacing=0.04, \n subplot_titles=[\"Device Browser - Visitors\", \"Device Browser - Paying Customers\", \"Device Browser - Mean Revenue\",\n \"Device Category - Visitors\", \"Device Category - Paying Customers\", \"Device Category - Mean Revenue\", \n \"Device OS - Visitors\", \"Device OS - Paying Customers\", \"Device OS - Mean Revenue\"])\n\nfig.append_trace(trace1, 1, 1)\nfig.append_trace(trace2, 1, 2)\nfig.append_trace(trace3, 1, 3)\nfig.append_trace(trace4, 2, 1)\nfig.append_trace(trace5, 2, 2)\nfig.append_trace(trace6, 2, 3)\nfig.append_trace(trace7, 3, 1)\nfig.append_trace(trace8, 3, 2)\nfig.append_trace(trace9, 3, 3)\n\nfig['layout'].update(height=1200, width=1200, paper_bgcolor='rgb(233,233,233)', title=\"Device Plots\")\ninit_notebook_mode(connected=True)\niplot(fig, filename='device-plots')","execution_count":55,"outputs":[{"output_type":"stream","text":"This is the format of your plot grid:\n[ (1,1) x1,y1 ] [ (1,2) x2,y2 ] [ (1,3) x3,y3 ]\n[ (2,1) x4,y4 ] [ (2,2) x5,y5 ] [ (2,3) x6,y6 ]\n[ (3,1) x7,y7 ] [ (3,2) x8,y8 ] [ (3,3) x9,y9 ]\n\n","name":"stdout"},{"output_type":"display_data","data":{"text/html":"","text/vnd.plotly.v1+html":""},"metadata":{}},{"output_type":"display_data","data":{"application/vnd.plotly.v1+json":{"data":[{"marker":{"color":"rgba(149, 48, 90, 0.6)"},"orientation":"h","showlegend":false,"x":[104,131,154,155,202,341,372,635,3036,11792],"y":["Opera","Safari (in-app)","Samsung Internet","Opera Mini","Edge","Android Webview","Internet Explorer","Firefox","Safari","Chrome"],"type":"bar","uid":"f77c32ba-66cc-49ca-9506-97456b38c4c2","xaxis":"x","yaxis":"y"},{"marker":{"color":"rgba(149, 48, 90, 0.6)"},"orientation":"h","showlegend":false,"x":[0,0,0,0,0,0,2,3,10,138],"y":["Opera","Safari (in-app)","Samsung Internet","Opera Mini","Edge","Android Webview","Internet Explorer","Firefox","Safari","Chrome"],"type":"bar","uid":"7a03c112-31cd-40bb-8df3-11ca80215d75","xaxis":"x2","yaxis":"y2"},{"marker":{"color":"rgba(149, 48, 90, 0.6)"},"orientation":"h","showlegend":false,"x":[null,null,null,null,null,null,37485000,129420000,11739000,131798115.94202898],"y":["Opera","Safari (in-app)","Samsung Internet","Opera Mini","Edge","Android Webview","Internet Explorer","Firefox","Safari","Chrome"],"type":"bar","uid":"8cd9a96c-7dfc-475f-ba99-1c382ffdda15","xaxis":"x3","yaxis":"y3"},{"marker":{"color":"rgba(128, 51, 89, 0.6)"},"orientation":"h","showlegend":false,"x":[642,4726,11715],"y":["tablet","mobile","desktop"],"type":"bar","uid":"014f3282-181e-4778-8c89-7a52d3f1657c","xaxis":"x4","yaxis":"y4"},{"marker":{"color":"rgba(128, 51, 89, 0.6)"},"orientation":"h","showlegend":false,"x":[2,10,141],"y":["tablet","mobile","desktop"],"type":"bar","uid":"d355396b-d8e9-44b9-8db0-15e01580b5cc","xaxis":"x5","yaxis":"y5"},{"marker":{"color":"rgba(128, 51, 89, 0.6)"},"orientation":"h","showlegend":false,"x":[22485000,23399000,131133333.33333333],"y":["tablet","mobile","desktop"],"type":"bar","uid":"21184e0a-a2eb-4ae9-92d7-999fb12c3156","xaxis":"x6","yaxis":"y6"},{"marker":{"color":"rgba(60, 150, 60, 0.6)"},"orientation":"h","showlegend":false,"x":[11,12,19,117,510,595,2111,3063,4363,6268],"y":["Tizen","Samsung","Windows Phone","(not set)","Chrome OS","Linux","iOS","Android","Macintosh","Windows"],"type":"bar","uid":"f11d6933-d713-4112-b54f-ef4bb3c3e39a","xaxis":"x7","yaxis":"y7"},{"marker":{"color":"rgba(60, 150, 60, 0.6)"},"orientation":"h","showlegend":false,"x":[0,0,0,0,11,9,6,6,96,25],"y":["Tizen","Samsung","Windows Phone","(not set)","Chrome OS","Linux","iOS","Android","Macintosh","Windows"],"type":"bar","uid":"8de6609a-3418-4d7f-afe0-3d02e726750f","xaxis":"x8","yaxis":"y8"},{"marker":{"color":"rgba(60, 150, 60, 0.6)"},"orientation":"h","showlegend":false,"x":[null,null,null,null,105913636.36363636,86060000,10708333.333333334,35785000,137164479.16666666,135296800],"y":["Tizen","Samsung","Windows Phone","(not set)","Chrome OS","Linux","iOS","Android","Macintosh","Windows"],"type":"bar","uid":"ed1a07f7-f87c-4c8d-b85d-2028eed2c614","xaxis":"x9","yaxis":"y9"}],"layout":{"annotations":[{"font":{"size":16},"showarrow":false,"text":"Device Browser - Visitors","x":0.14444444444444446,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Device Browser - Paying Customers","x":0.5,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Device Browser - Mean Revenue","x":0.8555555555555556,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Device Category - Visitors","x":0.14444444444444446,"xanchor":"center","xref":"paper","y":0.6533333333333333,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Device Category - Paying Customers","x":0.5,"xanchor":"center","xref":"paper","y":0.6533333333333333,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Device Category - Mean Revenue","x":0.8555555555555556,"xanchor":"center","xref":"paper","y":0.6533333333333333,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Device OS - Visitors","x":0.14444444444444446,"xanchor":"center","xref":"paper","y":0.3066666666666667,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Device OS - Paying Customers","x":0.5,"xanchor":"center","xref":"paper","y":0.3066666666666667,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Device OS - Mean Revenue","x":0.8555555555555556,"xanchor":"center","xref":"paper","y":0.3066666666666667,"yanchor":"bottom","yref":"paper"}],"xaxis":{"anchor":"y","domain":[0,0.2888888888888889]},"yaxis":{"anchor":"x","domain":[0.6933333333333334,1]},"xaxis2":{"anchor":"y2","domain":[0.35555555555555557,0.6444444444444445]},"yaxis2":{"anchor":"x2","domain":[0.6933333333333334,1]},"xaxis3":{"anchor":"y3","domain":[0.7111111111111111,1]},"yaxis3":{"anchor":"x3","domain":[0.6933333333333334,1]},"xaxis4":{"anchor":"y4","domain":[0,0.2888888888888889]},"yaxis4":{"anchor":"x4","domain":[0.3466666666666667,0.6533333333333333]},"xaxis5":{"anchor":"y5","domain":[0.35555555555555557,0.6444444444444445]},"yaxis5":{"anchor":"x5","domain":[0.3466666666666667,0.6533333333333333]},"xaxis6":{"anchor":"y6","domain":[0.7111111111111111,1]},"yaxis6":{"anchor":"x6","domain":[0.3466666666666667,0.6533333333333333]},"xaxis7":{"anchor":"y7","domain":[0,0.2888888888888889]},"yaxis7":{"anchor":"x7","domain":[0,0.3066666666666667]},"xaxis8":{"anchor":"y8","domain":[0.35555555555555557,0.6444444444444445]},"yaxis8":{"anchor":"x8","domain":[0,0.3066666666666667]},"xaxis9":{"anchor":"y9","domain":[0.7111111111111111,1]},"yaxis9":{"anchor":"x9","domain":[0,0.3066666666666667]},"height":1200,"width":1200,"paper_bgcolor":"rgb(233,233,233)","title":{"text":"Device Plots"}},"config":{"showLink":false,"linkText":"Export to plot.ly","plotlyServerURL":"https://plot.ly"}},"text/html":"
","text/vnd.plotly.v1+html":"
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Continent\ncnt_srs = datavis.groupby('geoNetwork.continent')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace1 = horizontal_bar_chart(cnt_srs[\"count\"].head(10), 'rgba(78, 50, 120, 0.6)')\ntrace2 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(10), 'rgba(78, 50, 120, 0.6)')\ntrace3 = horizontal_bar_chart(cnt_srs[\"mean\"].head(10), 'rgba(78, 50, 120, 0.6)')\n\n# Sub-continent\ncnt_srs = datavis.groupby('geoNetwork.subContinent')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace4 = horizontal_bar_chart(cnt_srs[\"count\"], 'black')\ntrace5 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"], 'black')\ntrace6 = horizontal_bar_chart(cnt_srs[\"mean\"], 'black')\n\n# Network domain\ncnt_srs = datavis.groupby('geoNetwork.networkDomain')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace7 = horizontal_bar_chart(cnt_srs[\"count\"].head(10), 'orange')\ntrace8 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(10), 'orange')\ntrace9 = horizontal_bar_chart(cnt_srs[\"mean\"].head(10), 'orange')\n\n# Creating two subplots\nfig = tools.make_subplots(rows=3, cols=3, vertical_spacing=0.08, horizontal_spacing=0.15, \n subplot_titles=[\"Continent - Visitors\", \"Continent - Paying Customers\", \"Continent - Mean Revenue\",\n \"Sub Continent - Visitors\", \"Sub Continent - Paying Customers\", \"Sub Continent - Mean Revenue\",\n \"Network Domain - Visitors\", \"Network Domain - Paying Customers\", \"Network Domain - Mean Revenue\"])\n\nfig.append_trace(trace1, 1, 1)\nfig.append_trace(trace2, 1, 2)\nfig.append_trace(trace3, 1, 3)\nfig.append_trace(trace4, 2, 1)\nfig.append_trace(trace5, 2, 2)\nfig.append_trace(trace6, 2, 3)\nfig.append_trace(trace7, 3, 1)\nfig.append_trace(trace8, 3, 2)\nfig.append_trace(trace9, 3, 3)\n\nfig['layout'].update(height=1500, width=1200, paper_bgcolor='rgb(233,233,233)', title=\"Geography Plots\")\ninit_notebook_mode(connected=True)\niplot(fig, filename='geo-plots')","execution_count":56,"outputs":[{"output_type":"stream","text":"This is the format of your plot grid:\n[ (1,1) x1,y1 ] [ (1,2) x2,y2 ] [ (1,3) x3,y3 ]\n[ (2,1) x4,y4 ] [ (2,2) x5,y5 ] [ (2,3) x6,y6 ]\n[ (3,1) x7,y7 ] [ (3,2) x8,y8 ] [ (3,3) x9,y9 ]\n\n","name":"stdout"},{"output_type":"display_data","data":{"text/html":"","text/vnd.plotly.v1+html":""},"metadata":{}},{"output_type":"display_data","data":{"application/vnd.plotly.v1+json":{"data":[{"marker":{"color":"rgba(78, 50, 120, 0.6)"},"orientation":"h","showlegend":false,"x":[28,280,380,3730,3931,8734],"y":["(not set)","Oceania","Africa","Europe","Asia","Americas"],"type":"bar","uid":"285d88b1-22f0-44f9-b83f-85f19ec8928c","xaxis":"x","yaxis":"y"},{"marker":{"color":"rgba(78, 50, 120, 0.6)"},"orientation":"h","showlegend":false,"x":[0,0,0,0,0,153],"y":["(not set)","Oceania","Africa","Europe","Asia","Americas"],"type":"bar","uid":"1cfeb293-a92a-4268-b0a0-29b7ae953142","xaxis":"x2","yaxis":"y2"},{"marker":{"color":"rgba(78, 50, 120, 0.6)"},"orientation":"h","showlegend":false,"x":[null,null,null,null,null,122671633.9869281],"y":["(not set)","Oceania","Africa","Europe","Asia","Americas"],"type":"bar","uid":"b86acfe5-45a9-41b5-960e-432b821aa9d8","xaxis":"x3","yaxis":"y3"},{"marker":{"color":"black"},"orientation":"h","showlegend":false,"x":[1,1,9,23,28,42,54,63,71,183,278,284,619,718,777,791,919,1104,1117,1176,1194,7631],"y":["Melanesia","Micronesian Region","Middle Africa","Central Asia","(not set)","Caribbean","Eastern Africa","Southern Africa","Western Africa","Northern Africa","Australasia","Central America","Western Asia","Southern Europe","South America","Eastern Europe","Eastern Asia","Northern Europe","Western Europe","Southern Asia","Southeast Asia","Northern America"],"type":"bar","uid":"db7248f0-4fd2-443f-a751-05ea1e141aa1","xaxis":"x4","yaxis":"y4"},{"marker":{"color":"black"},"orientation":"h","showlegend":false,"x":[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,152],"y":["Melanesia","Micronesian Region","Middle Africa","Central Asia","(not set)","Caribbean","Eastern Africa","Southern Africa","Western Africa","Northern Africa","Australasia","Central America","Western Asia","Southern Europe","South America","Eastern Europe","Eastern Asia","Northern Europe","Western Europe","Southern Asia","Southeast Asia","Northern America"],"type":"bar","uid":"9b444310-1d1c-4ee5-adab-41c276991612","xaxis":"x5","yaxis":"y5"},{"marker":{"color":"black"},"orientation":"h","showlegend":false,"x":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,97410000,null,null,null,null,null,null,122837828.94736843],"y":["Melanesia","Micronesian Region","Middle Africa","Central Asia","(not set)","Caribbean","Eastern Africa","Southern Africa","Western Africa","Northern Africa","Australasia","Central America","Western Asia","Southern Europe","South America","Eastern Europe","Eastern Asia","Northern Europe","Western Europe","Southern Asia","Southeast Asia","Northern America"],"type":"bar","uid":"d32ebb31-6ef6-4bb4-ac8a-843a90938cb4","xaxis":"x6","yaxis":"y6"},{"marker":{"color":"orange"},"orientation":"h","showlegend":false,"x":[123,125,151,168,177,260,282,564,2789,4907],"y":["cox.net","virginm.net","ttnet.com.tr","hinet.net","comcastbusiness.net","rr.com","verizon.net","comcast.net","unknown.unknown","(not set)"],"type":"bar","uid":"5753fafe-4058-4551-9a69-cef901f97f04","xaxis":"x7","yaxis":"y7"},{"marker":{"color":"orange"},"orientation":"h","showlegend":false,"x":[4,0,0,0,1,1,7,12,5,96],"y":["cox.net","virginm.net","ttnet.com.tr","hinet.net","comcastbusiness.net","rr.com","verizon.net","comcast.net","unknown.unknown","(not set)"],"type":"bar","uid":"95300157-8a85-4b9d-88cb-d2840978cfb5","xaxis":"x8","yaxis":"y8"},{"marker":{"color":"orange"},"orientation":"h","showlegend":false,"x":[212692500,null,null,null,79990000,1500000,87534285.71428572,63555000,36036000,137501458.33333334],"y":["cox.net","virginm.net","ttnet.com.tr","hinet.net","comcastbusiness.net","rr.com","verizon.net","comcast.net","unknown.unknown","(not set)"],"type":"bar","uid":"a26a4ead-1bc2-4532-b780-4a454533cf01","xaxis":"x9","yaxis":"y9"}],"layout":{"annotations":[{"font":{"size":16},"showarrow":false,"text":"Continent - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Continent - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Continent - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Sub Continent - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":0.6399999999999999,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Sub Continent - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":0.6399999999999999,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Sub Continent - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":0.6399999999999999,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Network Domain - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":0.27999999999999997,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Network Domain - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":0.27999999999999997,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Network Domain - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":0.27999999999999997,"yanchor":"bottom","yref":"paper"}],"xaxis":{"anchor":"y","domain":[0,0.2333333333333333]},"yaxis":{"anchor":"x","domain":[0.72,1]},"xaxis2":{"anchor":"y2","domain":[0.3833333333333333,0.6166666666666666]},"yaxis2":{"anchor":"x2","domain":[0.72,1]},"xaxis3":{"anchor":"y3","domain":[0.7666666666666666,0.9999999999999999]},"yaxis3":{"anchor":"x3","domain":[0.72,1]},"xaxis4":{"anchor":"y4","domain":[0,0.2333333333333333]},"yaxis4":{"anchor":"x4","domain":[0.36,0.6399999999999999]},"xaxis5":{"anchor":"y5","domain":[0.3833333333333333,0.6166666666666666]},"yaxis5":{"anchor":"x5","domain":[0.36,0.6399999999999999]},"xaxis6":{"anchor":"y6","domain":[0.7666666666666666,0.9999999999999999]},"yaxis6":{"anchor":"x6","domain":[0.36,0.6399999999999999]},"xaxis7":{"anchor":"y7","domain":[0,0.2333333333333333]},"yaxis7":{"anchor":"x7","domain":[0,0.27999999999999997]},"xaxis8":{"anchor":"y8","domain":[0.3833333333333333,0.6166666666666666]},"yaxis8":{"anchor":"x8","domain":[0,0.27999999999999997]},"xaxis9":{"anchor":"y9","domain":[0.7666666666666666,0.9999999999999999]},"yaxis9":{"anchor":"x9","domain":[0,0.27999999999999997]},"height":1500,"width":1200,"paper_bgcolor":"rgb(233,233,233)","title":{"text":"Geography Plots"}},"config":{"showLink":false,"linkText":"Export to plot.ly","plotlyServerURL":"https://plot.ly"}},"text/html":"
","text/vnd.plotly.v1+html":"
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"cnt_srs = datavis.groupby('totals.pageviews')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace1 = horizontal_bar_chart(cnt_srs[\"count\"].head(60), 'gray')\ntrace2 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(60), 'gray')\ntrace5 = horizontal_bar_chart(cnt_srs[\"mean\"].head(60), 'gray')\n\n# Hits\ncnt_srs = datavis.groupby('totals.hits')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", 'mean']\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace3 = horizontal_bar_chart(cnt_srs[\"count\"].head(60), 'blue')\ntrace4 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(60), 'blue')\ntrace6 = horizontal_bar_chart(cnt_srs[\"mean\"].head(60), 'blue')\n\n# Creating two subplots\nfig = tools.make_subplots(rows=2, cols=3, vertical_spacing=0.08, horizontal_spacing=0.15, \n subplot_titles=[\"Total Pageviews - Visitors\", \"Total Pageviews - Paying Customers\", \"Total Pageviews - Mean Revenue\",\n \"Total Hits - Visitors\", \"Total Hits - Paying Customers\", \"Total Hits - Mean Revenue\"])\n\nfig.append_trace(trace1, 1, 1)\nfig.append_trace(trace2, 1, 2)\nfig.append_trace(trace5, 1, 3)\nfig.append_trace(trace3, 2, 1)\nfig.append_trace(trace4, 2, 2)\nfig.append_trace(trace6, 2, 3)\n\nfig['layout'].update(height=1200, width=900, paper_bgcolor='rgb(233,233,233)', title=\"Visitor Profile Plots\")\ninit_notebook_mode(connected=True)\niplot(fig, filename='visitor-profile-plots')","execution_count":57,"outputs":[{"output_type":"stream","text":"This is the format of your plot grid:\n[ (1,1) x1,y1 ] [ (1,2) x2,y2 ] [ (1,3) x3,y3 ]\n[ (2,1) x4,y4 ] [ (2,2) x5,y5 ] [ (2,3) x6,y6 ]\n\n","name":"stdout"},{"output_type":"display_data","data":{"text/html":"","text/vnd.plotly.v1+html":""},"metadata":{}},{"output_type":"display_data","data":{"application/vnd.plotly.v1+json":{"data":[{"marker":{"color":"gray"},"orientation":"h","showlegend":false,"x":[2,2,2,2,2,2,2,2,2,2,2,2,2,2,3,3,3,4,4,5,5,6,6,7,8,9,10,10,11,11,14,16,16,18,19,22,25,29,33,34,36,41,60,80,83,89,92,110,117,182,209,226,293,400,471,630,889,1412,2430,8848],"y":["37","44","50","51","82","73","70","77","61","63","64","66","74","58","52","49","46","38","42","48","40","36","43","35","39","34","32","41","33","30","28","25","29","31","27","26","24","22","19","20","21","23","18","16","17","15","14","12","13","11","10","9","8","7","6","5","4","3","2","1"],"type":"bar","uid":"a4d3301e-30ef-497d-8d78-51082861cc80","xaxis":"x","yaxis":"y"},{"marker":{"color":"gray"},"orientation":"h","showlegend":false,"x":[0,2,2,0,1,1,1,1,0,0,0,1,1,0,0,1,0,1,1,2,0,1,1,1,2,4,5,3,2,2,7,4,2,2,3,3,3,3,4,5,2,6,4,8,9,9,7,5,2,11,5,3,1,2,1,0,0,0,0,0],"y":["37","44","50","51","82","73","70","77","61","63","64","66","74","58","52","49","46","38","42","48","40","36","43","35","39","34","32","41","33","30","28","25","29","31","27","26","24","22","19","20","21","23","18","16","17","15","14","12","13","11","10","9","8","7","6","5","4","3","2","1"],"type":"bar","uid":"7d6c187c-3633-433d-a3b6-d115328564e4","xaxis":"x2","yaxis":"y2"},{"marker":{"color":"gray"},"orientation":"h","showlegend":false,"x":[null,241485000,95945000,null,56910000,52450000,3990000,352290000,null,null,null,59900000,17590000,null,null,22990000,null,97410000,271580000,58020000,null,81580000,86340000,37990000,289705000,230275000,58340000,41380000,37365000,159480000,88921428.57142857,215547500,186155000,110640000,178496666.66666666,81500000,117673333.33333333,248203333.33333334,32105000,53790000,418075000,56318333.333333336,321390000,102653750,46022222.222222224,40523333.333333336,62754285.71428572,94048000,33585000,40844545.45454545,33576000,61280000,27590000,40090000,17590000,null,null,null,null,null],"y":["37","44","50","51","82","73","70","77","61","63","64","66","74","58","52","49","46","38","42","48","40","36","43","35","39","34","32","41","33","30","28","25","29","31","27","26","24","22","19","20","21","23","18","16","17","15","14","12","13","11","10","9","8","7","6","5","4","3","2","1"],"type":"bar","uid":"991258f3-d34c-4111-bd38-5570e425d68d","xaxis":"x3","yaxis":"y3"},{"marker":{"color":"blue"},"orientation":"h","showlegend":false,"x":[2,3,3,3,3,3,4,4,4,5,5,5,5,5,6,6,7,8,8,10,10,11,11,12,14,15,16,17,18,21,22,22,23,26,26,40,40,43,44,46,59,61,61,69,89,92,124,127,134,172,211,261,304,387,454,636,816,1325,2335,8726],"y":["58","57","62","63","54","52","40","66","45","61","50","48","55","49","51","47","46","42","44","33","38","43","35","41","39","36","37","34","31","29","32","30","28","26","27","23","25","24","20","22","19","21","18","17","16","15","14","13","12","11","10","9","8","7","6","5","4","3","2","1"],"type":"bar","uid":"af26899d-755f-4e85-8e02-5a3bf70bd0dc","xaxis":"x4","yaxis":"y4"},{"marker":{"color":"blue"},"orientation":"h","showlegend":false,"x":[2,0,0,1,0,1,2,3,3,2,2,1,0,0,1,0,1,0,1,1,2,4,1,2,2,2,3,4,3,0,3,4,2,2,3,3,5,4,3,3,6,6,5,6,7,3,4,6,7,4,2,1,1,2,0,0,0,0,0,0],"y":["58","57","62","63","54","52","40","66","45","61","50","48","55","49","51","47","46","42","44","33","38","43","35","41","39","36","37","34","31","29","32","30","28","26","27","23","25","24","20","22","19","21","18","17","16","15","14","13","12","11","10","9","8","7","6","5","4","3","2","1"],"type":"bar","uid":"23a61e03-91fa-4aa6-be03-e2a6645d8a1a","xaxis":"x5","yaxis":"y5"},{"marker":{"color":"blue"},"orientation":"h","showlegend":false,"x":[25895000,null,null,57560000,null,394230000,245205000,83456666.66666667,106456666.66666667,180160000,436055000,21980000,null,null,37990000,null,97410000,null,18460000,71850000,64375000,149395000,64760000,45570000,18175000,114300000,166236666.66666666,63742500,76373333.33333333,null,70173333.33333333,290230000,43085000,15490000,436146666.6666667,167486666.66666666,237568000,47980000,23920000,13856666.666666666,18495000,69470000,164688000,47928333.333333336,29815714.285714287,31750000,26592500,49191666.666666664,90932857.14285715,29520000,30745000,41590000,27590000,38385000,null,null,null,null,null,null],"y":["58","57","62","63","54","52","40","66","45","61","50","48","55","49","51","47","46","42","44","33","38","43","35","41","39","36","37","34","31","29","32","30","28","26","27","23","25","24","20","22","19","21","18","17","16","15","14","13","12","11","10","9","8","7","6","5","4","3","2","1"],"type":"bar","uid":"b4089709-e85a-4335-a490-2c45270b15c6","xaxis":"x6","yaxis":"y6"}],"layout":{"annotations":[{"font":{"size":16},"showarrow":false,"text":"Total Pageviews - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Total Pageviews - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Total Pageviews - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Total Hits - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Total Hits - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Total Hits - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"}],"xaxis":{"anchor":"y","domain":[0,0.2333333333333333]},"yaxis":{"anchor":"x","domain":[0.54,1]},"xaxis2":{"anchor":"y2","domain":[0.3833333333333333,0.6166666666666666]},"yaxis2":{"anchor":"x2","domain":[0.54,1]},"xaxis3":{"anchor":"y3","domain":[0.7666666666666666,0.9999999999999999]},"yaxis3":{"anchor":"x3","domain":[0.54,1]},"xaxis4":{"anchor":"y4","domain":[0,0.2333333333333333]},"yaxis4":{"anchor":"x4","domain":[0,0.46]},"xaxis5":{"anchor":"y5","domain":[0.3833333333333333,0.6166666666666666]},"yaxis5":{"anchor":"x5","domain":[0,0.46]},"xaxis6":{"anchor":"y6","domain":[0.7666666666666666,0.9999999999999999]},"yaxis6":{"anchor":"x6","domain":[0,0.46]},"height":1200,"width":900,"paper_bgcolor":"rgb(233,233,233)","title":{"text":"Visitor Profile Plots"}},"config":{"showLink":false,"linkText":"Export to plot.ly","plotlyServerURL":"https://plot.ly"}},"text/html":"
","text/vnd.plotly.v1+html":"
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"cnt_srs = datavis.groupby('trafficSource.source')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace1 = horizontal_bar_chart(cnt_srs[\"count\"].head(10), 'orange')\ntrace2 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(10), 'orange')\ntrace3 = horizontal_bar_chart(cnt_srs[\"mean\"].head(10), 'orange')\n\n# Sub-continent\ncnt_srs = datavis.groupby('trafficSource.medium')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace4 = horizontal_bar_chart(cnt_srs[\"count\"], 'green')\ntrace5 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"], 'green')\ntrace6 = horizontal_bar_chart(cnt_srs[\"mean\"], 'green')\n\n# Creating two subplots\nfig = tools.make_subplots(rows=2, cols=3, vertical_spacing=0.08, horizontal_spacing=0.15, \n subplot_titles=[\"Traffic Source - Visitors\", \"Traffic Source - Paying Customers\", \"Traffic Source - Mean Revenue\",\n \"Traffic Source Medium - Visitors\", \"Traffic Source Medium - Paying Customers\", \"Traffic Source Medium - Mean Revenue\"\n ])\n\nfig.append_trace(trace1, 1, 1)\nfig.append_trace(trace2, 1, 2)\nfig.append_trace(trace3, 1, 3)\nfig.append_trace(trace4, 2, 1)\nfig.append_trace(trace5, 2, 2)\nfig.append_trace(trace6, 2, 3)\n\nfig['layout'].update(height=1000, width=1200, paper_bgcolor='rgb(233,233,233)', title=\"Traffic Source Plots\")\ninit_notebook_mode(connected=True)\niplot(fig, filename='traffic-source-plots')","execution_count":58,"outputs":[{"output_type":"stream","text":"This is the format of your plot grid:\n[ (1,1) x1,y1 ] [ (1,2) x2,y2 ] [ (1,3) x3,y3 ]\n[ (2,1) x4,y4 ] [ (2,2) x5,y5 ] [ (2,3) x6,y6 ]\n\n","name":"stdout"},{"output_type":"display_data","data":{"text/html":"","text/vnd.plotly.v1+html":""},"metadata":{}},{"output_type":"display_data","data":{"application/vnd.plotly.v1+json":{"data":[{"marker":{"color":"orange"},"orientation":"h","showlegend":false,"x":[43,63,73,82,86,326,377,3240,5696,6582],"y":["reddit.com","m.facebook.com","dfa","google.com","sites.google.com","Partners","analytics.google.com","youtube.com","(direct)","google"],"type":"bar","uid":"3585a473-1ed7-4e46-b96a-4d0426a8b0e9","xaxis":"x","yaxis":"y"},{"marker":{"color":"orange"},"orientation":"h","showlegend":false,"x":[0,0,0,1,0,0,0,1,112,37],"y":["reddit.com","m.facebook.com","dfa","google.com","sites.google.com","Partners","analytics.google.com","youtube.com","(direct)","google"],"type":"bar","uid":"83e587fd-d082-4db0-8fd9-0983d50de315","xaxis":"x2","yaxis":"y2"},{"marker":{"color":"orange"},"orientation":"h","showlegend":false,"x":[null,null,null,17940000,null,null,null,5000000,129706428.57142857,92860000],"y":["reddit.com","m.facebook.com","dfa","google.com","sites.google.com","Partners","analytics.google.com","youtube.com","(direct)","google"],"type":"bar","uid":"76d29b01-88c8-42ab-baf8-2678b8a03678","xaxis":"x3","yaxis":"y3"},{"marker":{"color":"green"},"orientation":"h","showlegend":false,"x":[94,326,754,4303,5695,5911],"y":["cpm","affiliate","cpc","referral","(none)","organic"],"type":"bar","uid":"ecad0fb4-a1b9-41da-8403-df5c0c10f9f2","xaxis":"x4","yaxis":"y4"},{"marker":{"color":"green"},"orientation":"h","showlegend":false,"x":[0,0,4,4,112,33],"y":["cpm","affiliate","cpc","referral","(none)","organic"],"type":"bar","uid":"85da3472-1c46-4ff3-bf25-9f285454f1fa","xaxis":"x5","yaxis":"y5"},{"marker":{"color":"green"},"orientation":"h","showlegend":false,"x":[null,null,213630000,201455000,129706428.57142857,78221212.12121212],"y":["cpm","affiliate","cpc","referral","(none)","organic"],"type":"bar","uid":"d08fc551-f596-439c-a781-d3e62da9e7c8","xaxis":"x6","yaxis":"y6"}],"layout":{"annotations":[{"font":{"size":16},"showarrow":false,"text":"Traffic Source - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Traffic Source - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Traffic Source - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Traffic Source Medium - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Traffic Source Medium - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Traffic Source Medium - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"}],"xaxis":{"anchor":"y","domain":[0,0.2333333333333333]},"yaxis":{"anchor":"x","domain":[0.54,1]},"xaxis2":{"anchor":"y2","domain":[0.3833333333333333,0.6166666666666666]},"yaxis2":{"anchor":"x2","domain":[0.54,1]},"xaxis3":{"anchor":"y3","domain":[0.7666666666666666,0.9999999999999999]},"yaxis3":{"anchor":"x3","domain":[0.54,1]},"xaxis4":{"anchor":"y4","domain":[0,0.2333333333333333]},"yaxis4":{"anchor":"x4","domain":[0,0.46]},"xaxis5":{"anchor":"y5","domain":[0.3833333333333333,0.6166666666666666]},"yaxis5":{"anchor":"x5","domain":[0,0.46]},"xaxis6":{"anchor":"y6","domain":[0.7666666666666666,0.9999999999999999]},"yaxis6":{"anchor":"x6","domain":[0,0.46]},"height":1000,"width":1200,"paper_bgcolor":"rgb(233,233,233)","title":{"text":"Traffic Source Plots"}},"config":{"showLink":false,"linkText":"Export to plot.ly","plotlyServerURL":"https://plot.ly"}},"text/html":"
","text/vnd.plotly.v1+html":"
"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"cnt_srs = datavis.groupby('totals.sessionQualityDim')['totals.transactionRevenue'].agg(['size', 'count', 'mean'])\ncnt_srs.columns = [\"count\", \"count of non-zero revenue\", \"mean\"]\ncnt_srs = cnt_srs.sort_values(by=\"count\", ascending=False)\ntrace1 = horizontal_bar_chart(cnt_srs[\"count\"].head(60), 'cyan')\ntrace2 = horizontal_bar_chart(cnt_srs[\"count of non-zero revenue\"].head(60), 'cyan')\ntrace5 = horizontal_bar_chart(cnt_srs[\"mean\"].head(60), 'cyan')\n\n# Creating two subplots\nfig = tools.make_subplots(rows=2, cols=3, vertical_spacing=0.08, horizontal_spacing=0.15, \n subplot_titles=[\"Session Quality - Visitors\", \"Session Quality - Paying Customers\", \"Session Quality - Mean Revenue\",\n \"Time On Site - Visitors\", \"Time On Site - Paying Customers\", \"Time On Site - Mean Revenue\"])\n\nfig.append_trace(trace1, 1, 1)\nfig.append_trace(trace2, 1, 2)\nfig.append_trace(trace5, 1, 3)\nfig.append_trace(trace3, 2, 1)\nfig.append_trace(trace4, 2, 2)\nfig.append_trace(trace6, 2, 3)\n\nfig['layout'].update(height=1200, width=900, paper_bgcolor='rgb(233,233,233)', title=\"Visitor Profile Plots\")\ninit_notebook_mode(connected=True)\niplot(fig, filename='visitor-profile-plots')","execution_count":59,"outputs":[{"output_type":"stream","text":"This is the format of your plot grid:\n[ (1,1) x1,y1 ] [ (1,2) x2,y2 ] [ (1,3) x3,y3 ]\n[ (2,1) x4,y4 ] [ (2,2) x5,y5 ] [ (2,3) x6,y6 ]\n\n","name":"stdout"},{"output_type":"display_data","data":{"text/html":"","text/vnd.plotly.v1+html":""},"metadata":{}},{"output_type":"display_data","data":{"application/vnd.plotly.v1+json":{"data":[{"marker":{"color":"cyan"},"orientation":"h","showlegend":false,"x":[5,5,5,5,5,5,5,5,5,6,6,6,6,6,6,6,6,6,6,7,7,7,7,7,7,7,7,8,8,8,8,8,8,8,8,9,9,9,9,9,9,10,10,10,12,13,13,17,22,25,25,28,30,42,51,72,107,157,491,7263],"y":["48","85","64","65","18","78","81","84","40","34","49","46","30","50","52","80","57","59","67","75","26","29","42","63","71","33","38","24","28","51","36","66","39","56","54","17","20","53","23","22","21","19","45","25","15","16","14","12","13","11","8","9","10","7","6","5","4","3","2","1"],"type":"bar","uid":"1c2a1a71-0348-4fc9-9d3b-5fb60df083aa","xaxis":"x","yaxis":"y"},{"marker":{"color":"cyan"},"orientation":"h","showlegend":false,"x":[0,3,0,1,0,3,3,5,0,0,1,0,0,1,0,2,1,1,3,1,0,0,0,3,2,0,0,0,0,1,0,0,0,0,2,0,0,1,1,0,0,0,2,0,0,0,1,0,0,0,1,0,0,0,0,0,0,1,0,0],"y":["48","85","64","65","18","78","81","84","40","34","49","46","30","50","52","80","57","59","67","75","26","29","42","63","71","33","38","24","28","51","36","66","39","56","54","17","20","53","23","22","21","19","45","25","15","16","14","12","13","11","8","9","10","7","6","5","4","3","2","1"],"type":"bar","uid":"f4c1117c-002e-4a2d-a402-4214cd066652","xaxis":"x2","yaxis":"y2"},{"marker":{"color":"cyan"},"orientation":"h","showlegend":false,"x":[null,60386666.666666664,null,21990000,null,187410000,88233333.33333333,131982000,null,null,5000000,null,null,21000000,null,25760000,58480000,1227800000,24643333.333333332,79990000,null,null,null,50583333.333333336,19460000,null,null,null,null,59180000,null,null,null,null,101240000,null,null,17590000,21980000,null,null,null,66150000,null,null,null,41590000,null,null,null,27590000,null,null,null,null,null,null,17590000,null,null],"y":["48","85","64","65","18","78","81","84","40","34","49","46","30","50","52","80","57","59","67","75","26","29","42","63","71","33","38","24","28","51","36","66","39","56","54","17","20","53","23","22","21","19","45","25","15","16","14","12","13","11","8","9","10","7","6","5","4","3","2","1"],"type":"bar","uid":"ca421361-a025-43b0-bd51-5029cca729b3","xaxis":"x3","yaxis":"y3"},{"marker":{"color":"orange"},"orientation":"h","showlegend":false,"x":[null,null,null,17940000,null,null,null,5000000,129706428.57142857,92860000],"y":["reddit.com","m.facebook.com","dfa","google.com","sites.google.com","Partners","analytics.google.com","youtube.com","(direct)","google"],"type":"bar","uid":"4a946f60-0e65-4f56-adfb-0bb395e00a21","xaxis":"x4","yaxis":"y4"},{"marker":{"color":"green"},"orientation":"h","showlegend":false,"x":[94,326,754,4303,5695,5911],"y":["cpm","affiliate","cpc","referral","(none)","organic"],"type":"bar","uid":"10d33b34-4002-484c-bc38-cd25f7b208a7","xaxis":"x5","yaxis":"y5"},{"marker":{"color":"green"},"orientation":"h","showlegend":false,"x":[null,null,213630000,201455000,129706428.57142857,78221212.12121212],"y":["cpm","affiliate","cpc","referral","(none)","organic"],"type":"bar","uid":"769cbe34-cca9-43ff-a5aa-5948aff860c1","xaxis":"x6","yaxis":"y6"}],"layout":{"annotations":[{"font":{"size":16},"showarrow":false,"text":"Session Quality - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Session Quality - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Session Quality - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":1,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Time On Site - Visitors","x":0.11666666666666665,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Time On Site - Paying Customers","x":0.49999999999999994,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"},{"font":{"size":16},"showarrow":false,"text":"Time On Site - Mean Revenue","x":0.8833333333333333,"xanchor":"center","xref":"paper","y":0.46,"yanchor":"bottom","yref":"paper"}],"xaxis":{"anchor":"y","domain":[0,0.2333333333333333]},"yaxis":{"anchor":"x","domain":[0.54,1]},"xaxis2":{"anchor":"y2","domain":[0.3833333333333333,0.6166666666666666]},"yaxis2":{"anchor":"x2","domain":[0.54,1]},"xaxis3":{"anchor":"y3","domain":[0.7666666666666666,0.9999999999999999]},"yaxis3":{"anchor":"x3","domain":[0.54,1]},"xaxis4":{"anchor":"y4","domain":[0,0.2333333333333333]},"yaxis4":{"anchor":"x4","domain":[0,0.46]},"xaxis5":{"anchor":"y5","domain":[0.3833333333333333,0.6166666666666666]},"yaxis5":{"anchor":"x5","domain":[0,0.46]},"xaxis6":{"anchor":"y6","domain":[0.7666666666666666,0.9999999999999999]},"yaxis6":{"anchor":"x6","domain":[0,0.46]},"height":1200,"width":900,"paper_bgcolor":"rgb(233,233,233)","title":{"text":"Visitor Profile Plots"}},"config":{"showLink":false,"linkText":"Export to plot.ly","plotlyServerURL":"https://plot.ly"}},"text/html":"
","text/vnd.plotly.v1+html":"
"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"\n### Feature Engineering"},{"metadata":{},"cell_type":"markdown","source":"The feature engineering function accomplishes numerous tasks necessary prior to the machine learning models. First, when analyzing the dataframe, it was determined that there were multiple features that were similar to the target feature totals.transactionRevenue such as totals.totalTransactionRevenue and totals.transactions. Including these in our analysis would likely have skewed the final results of the model as they contain very similar values to the target. Therefore, these features were removed. Other features deemed to possess no analytical value such as the visitID and hits were removed. Included in this category are features that have over 95% missing values such as trafficSource.adwordsClickInfo.adNetworkType. Values could have been imputed for the missing values in this columns, but when over 95% of the values in the column are missing it likely would not have yielded ideal results. \n\nOther columns with missing values were left unaddressed, as both of the models being employed in this problem are able to deal with missing values with strong in-built methods. Next, the categoric features were encoded, and the numeric columns were converted to float values. Additionally, missing values for the target totals.transactionRevenue were imputed with zero. This is reasonable, as missing values in these columns likely indicates that no transaction took place. Lastly, the date feature was converted into a standard date format. "},{"metadata":{"_uuid":"d2266fb9edd72541fd684879ba8f5f09654609a0","trusted":true},"cell_type":"code","source":"def featureEngineering(df):\n 'Function requires a dataframe as an input, and will output a feature Engineered dataframe'\n #Calculate number of constant columns that can be dropped\n columns_constant = [i for i in df.columns if df[i].nunique(dropna=False)==1 ]\n # Drop features that are very similar to our target feature or will have no analytical value\n similarToFeature = ['totals.totalTransactionRevenue'] + ['totals.transactions']\n noAnalyticalValue = ['visitId']+['customDimensions']+['hits']+['trafficSource.adwordsClickInfo.adNetworkType']+['trafficSource.adwordsClickInfo.isVideoAd']+['trafficSource.adwordsClickInfo.page']+['trafficSource.adwordsClickInfo.slot']\n tot = similarToFeature + noAnalyticalValue\n if 'totals.totalTransactionRevenue' and 'totals.transactions' and 'visitId' and 'fullVisitorId' and 'customDimensions' and 'hits' and 'trafficSource.adwordsClickInfo.adNetworkType' and 'trafficSource.adwordsClickInfo.isVideoAd'and 'trafficSource.adwordsClickInfo.page' and 'trafficSource.adwordsClickInfo.slot'in df.columns:\n columns_drop = columns_constant + tot\n #Drop constant columns \n df= df.drop(columns_drop, axis=1)\n else:\n df= df.drop(columns_constant, axis=1)\n \n #Encode the categoric features to numeric \n categoric_features = [\"channelGrouping\", \"device.browser\", \n \"device.deviceCategory\", \"device.operatingSystem\", \n \"geoNetwork.city\", \"geoNetwork.continent\", \n \"geoNetwork.country\", \"geoNetwork.metro\",\n \"geoNetwork.networkDomain\", \"geoNetwork.region\", \n \"geoNetwork.subContinent\", \"trafficSource.adContent\", \n \"trafficSource.adwordsClickInfo.adNetworkType\", \n \"trafficSource.adwordsClickInfo.gclId\", \n \"trafficSource.adwordsClickInfo.page\", \n \"trafficSource.adwordsClickInfo.slot\", \"trafficSource.campaign\",\n \"trafficSource.keyword\", \"trafficSource.medium\", \n \"trafficSource.referralPath\", \"trafficSource.source\",\n 'trafficSource.adwordsClickInfo.isVideoAd', 'trafficSource.isTrueDirect', 'device.isMobile']\n for i in categoric_features:\n if i in df.columns:\n encode = sklearn.preprocessing.LabelEncoder()\n encode.fit(df[i].values.astype('str'))\n df[i] = encode.transform(list(df[i].values.astype('str')))\n \n #Convert numeric features to float values\n numeric_features = [\"totals.hits\", \"totals.pageviews\",'totals.transactionRevenue',\"visitNumber\", \"visitStartTime\", 'totals.bounces', 'totals.newVisits','totals.sessionQualityDim', 'totals.timeOnSite']\n for i in numeric_features:\n df[i] = df[i].astype(float)\n \n #Encode Zeroes for nan values in totals.bounces & totals.transactionRevenues\n df['totals.transactionRevenue'].fillna(0,inplace=True)\n \n #Convert date to datetime\n df['date'] = pd.to_datetime(df['date'], format = \"%Y%m%d\")\n return df","execution_count":60,"outputs":[]},{"metadata":{"_uuid":"b4699c09da136c3af47aa8eab59c81a3e8b1696b","trusted":true},"cell_type":"code","source":"#FeatureEngineer training and testing data \ntrainingData = featureEngineering(trainingData)\ntestingData = featureEngineering(testingData)","execution_count":61,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#Attempted Reading in Batches\n# Function runs and produces an output but not as intended \n# def dataPipeline(filename):\n# 'dataPipeline is a function that takes a csv as an input, and returns a fully pre-processed Pandas dataframe ready for'\n# 'ML Algorithms'\n# #Start Time of Function\n# start_time = dt.datetime.now()\n# print(\"Started at \", start_time)\n# # Read number of file lines in the dataset\n# num_lines = int(sum(1 for line in open(filename))) \n# # Process in chunk sizes of 1%\n# batchSize = int(0.01*num_lines)\n# # Initialize a List to hold all of the processed, chunked dataframes\n# result = []\n# #For Parser\n# JSON_COLUMNS = ['device', 'geoNetwork', 'totals', 'trafficSource']\n# print('Entering into processing the csv file by batches')\n# for batch in pd.read_csv(filename, chunksize=batchSize,\n# converters={column: json.loads for column in JSON_COLUMNS}, \n# dtype={'fullVisitorId': 'str'}):\n# parsed = Parser(batch)\n# processed = featureEngineering(parsed)\n# result.append(processed)\n# Output = pd.concat(result)\n# #Time it took for function to finish\n# end_time = dt.datetime.now()\n# elapsed_time = end_time - start_time\n# print (\"Preprocessed \", filename, \"completely. Elapsed time: \", elapsed_time)\n# return Output","execution_count":62,"outputs":[]},{"metadata":{"_uuid":"089616ce9679f16c49696249f2a760e10224b480","trusted":true},"cell_type":"code","source":"print(trainingData.shape)\ntrainingData.head()","execution_count":63,"outputs":[{"output_type":"stream","text":"(17083, 31)\n","name":"stdout"},{"output_type":"execute_result","execution_count":63,"data":{"text/plain":" channelGrouping ... trafficSource.source\n0 1 ... 0\n1 3 ... 18\n2 0 ... 1\n3 3 ... 8\n4 3 ... 18\n\n[5 rows x 31 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
channelGroupingdatefullVisitorIdvisitNumbervisitStartTimedevice.browserdevice.deviceCategorydevice.isMobiledevice.operatingSystemgeoNetwork.citygeoNetwork.continentgeoNetwork.countrygeoNetwork.metrogeoNetwork.networkDomaingeoNetwork.regiongeoNetwork.subContinenttotals.bouncestotals.hitstotals.newVisitstotals.pageviewstotals.sessionQualityDimtotals.timeOnSitetotals.transactionRevenuetrafficSource.adContenttrafficSource.adwordsClickInfo.gclIdtrafficSource.campaigntrafficSource.isTrueDirecttrafficSource.keywordtrafficSource.mediumtrafficSource.referralPathtrafficSource.source
012017-10-1644475697593702303451.01.508203e+095111391215366158725112NaN4.01.04.01.066.00.025745309404660
132017-10-1684497279793338417971.01.508193e+09500634221535904312NaN6.01.06.02.040.00.025745313446618
202017-10-1602585609667935482231.01.508164e+0990015391410566110125121NaN7.01.07.01.0243.00.0257451019414661
332017-10-1621201296082331011231.01.508177e+09900153912956621312513NaN8.01.08.01.0111.00.025745315744668
432017-10-1638485238647587192061.01.508189e+095003391215366025112NaN9.01.07.06.0250.00.025745313446618
\n
"},"metadata":{}}]},{"metadata":{"_uuid":"8b00ba61a56bc25dfdf1f85817a146d7c7934a33","trusted":true},"cell_type":"code","source":"print(testingData.shape)\ntestingData.head()","execution_count":64,"outputs":[{"output_type":"stream","text":"(4015, 31)\n","name":"stdout"},{"output_type":"execute_result","execution_count":64,"data":{"text/plain":" channelGrouping ... trafficSource.source\n0 5 ... 0\n1 5 ... 0\n2 3 ... 12\n3 3 ... 12\n4 1 ... 0\n\n[5 rows x 31 columns]","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
channelGroupingdatefullVisitorIdvisitNumbervisitStartTimedevice.browserdevice.deviceCategorydevice.isMobiledevice.operatingSystemgeoNetwork.citygeoNetwork.continentgeoNetwork.countrygeoNetwork.metrogeoNetwork.networkDomaingeoNetwork.regiongeoNetwork.subContinenttotals.bouncestotals.hitstotals.newVisitstotals.pageviewstotals.sessionQualityDimtotals.timeOnSitetotals.transactionRevenuetrafficSource.adContenttrafficSource.adwordsClickInfo.gclIdtrafficSource.campaigntrafficSource.isTrueDirecttrafficSource.keywordtrafficSource.mediumtrafficSource.referralPathtrafficSource.source
052018-05-1152998207948451745082.01.526088e+09100392210538013810NaN5.0NaN3.01.028.00.001071131010
152018-05-1133265401216892397421.01.526068e+091003921051107610NaN5.01.04.01.042.00.001071131010
232018-05-1101214139715331760851.01.526098e+0910118672105371752410NaN6.01.04.01.060.00.001071114012
332018-05-1188166039955075506902.01.526078e+09100216421053702410NaN6.0NaN5.08.0258.00.0010710314012
412018-05-1162162041010749014301.01.526050e+09100416421053702410NaN14.01.012.052.0387.00.00107102000
\n
"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"\n## Building 2 Baseline Models"},{"metadata":{},"cell_type":"markdown","source":"The two baseline models chosen to be used for this regression task were the LGBM and the XGB models. The XGB model was chosen as it has proven to consistently perform well on machine learning challenges on Kaggle. The LGBM is a newer model that is less proven, but is stated to run multiple times faster than the XGB and achieve the same accuracies. The two models will be compared in this section. "},{"metadata":{},"cell_type":"markdown","source":"\n### Seperate Features, Target, and Construct Validation Set"},{"metadata":{},"cell_type":"markdown","source":"The first step taken in building the two models was the separation of the training set into a training and validation set based on an 80/20 split on the dat feature in the training set between the two. This split between the training and validation set is a typical sized split that should help prevent overtraining of the models being built. Afterwards, the three datasets were split into their respective targets and features. "},{"metadata":{"_uuid":"349497f1d6d4ff558288b554feccc6fed1d7dcf0","trusted":true},"cell_type":"code","source":"#Split training set for validation set \ntrain_df = trainingData[trainingData['date']> dt.date(2016,11,30)]\nvalidation_df = trainingData[trainingData['date']<=dt.date(2016,11,30)]\n\nfeatures = [] \nlabel = []\nfor i in trainingData.columns:\n if i != 'totals.transactionRevenue' and i!='date' and i!= 'fullVisitorId' and i!= 'visitStartTime':\n features.append(i)\n elif i == 'totals.transactionRevenue':\n label.append(i)\n#Target\ntrain_y = np.log1p(train_df[label])\nvalidation_y = np.log1p(validation_df[label])\ntest_y = np.log1p(testingData[label])\n#Features\ntrain_x = train_df[features]\nvalidation_x = validation_df[features]\ntest_x = testingData[features]","execution_count":65,"outputs":[{"output_type":"stream","text":"/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:2: FutureWarning:\n\nComparing Series of datetimes with 'datetime.date'. Currently, the\n'datetime.date' is coerced to a datetime. In the future pandas will\nnot coerce, and a TypeError will be raised. To retain the current\nbehavior, convert the 'datetime.date' to a datetime with\n'pd.Timestamp'.\n\n/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:3: FutureWarning:\n\nComparing Series of datetimes with 'datetime.date'. Currently, the\n'datetime.date' is coerced to a datetime. In the future pandas will\nnot coerce, and a TypeError will be raised. To retain the current\nbehavior, convert the 'datetime.date' to a datetime with\n'pd.Timestamp'.\n\n","name":"stderr"}]},{"metadata":{"_uuid":"aa982ce8c8b85cb02c57928678fe4059c3722e6d","trusted":true},"cell_type":"code","source":"print('training dataframe shape:' , train_df.shape)\nprint('validation dataframe shape:' , validation_df.shape)\nprint('training features shape:' , train_x.shape)\nprint('validation features shape:' , validation_x.shape)\nprint('training label shape:' , train_y.shape)\nprint('validation label shape:' , validation_y.shape)\nprint('testing features shape:' , test_x.shape)\nprint('testing label shape:' , test_y.shape)","execution_count":66,"outputs":[{"output_type":"stream","text":"training dataframe shape: (13617, 31)\nvalidation dataframe shape: (3466, 31)\ntraining features shape: (13617, 27)\nvalidation features shape: (3466, 27)\ntraining label shape: (13617, 1)\nvalidation label shape: (3466, 1)\ntesting features shape: (4015, 27)\ntesting label shape: (4015, 1)\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"\n\n### LGBM"},{"metadata":{},"cell_type":"markdown","source":"#### Hyperparameter Tuning LGBM"},{"metadata":{},"cell_type":"markdown","source":"The first step completed in constructing the LGBM model was hyperparameter tuning conducted using the the training set with the RandomizedSearchCV. The RandomizedSearchCV was used for hyperparameter tuning because it produces reasonable results in a very flexible and efficient manner. During hyperparameter tuning, this model ran extremely fast, being able to fit 3 cross validation folds with 100 iterations totalling 300 possible fits of hyperparameters. An important note here is that since early stopping is being used to optimize the number of estimators, one of the most important parameters in the model, n_estimators was set to an upper maximum bound. The final hyperparameters, other than n_estimators, can be found in the output below. The final tuned parameters were saved and can be seen in the dictionary \"SavedLGBMParams\" in the interest of efficiency."},{"metadata":{"_uuid":"a21e1f2475f595dffa803d2af7cc7bb802f96a7d","trusted":true},"cell_type":"code","source":"# Prior to early stopping, determine best hyperparameters for most important parameters in \n#LGBM models. n_estimators, one of the most important parameters, is determined by \n# early stopping, so a maximum limit is set here \nparams_LGB ={'num_leaves': sp_randint(6, 50), \n \"learning_rate\": sp_uniform(0.05, 0.4),\n 'min_child_samples': sp_randint(100, 500), \n 'min_child_weight': [1e-5, 1e-3, 1e-2, 1e-1, 1, 1e1, 1e2, 1e3, 1e4],\n 'subsample': sp_uniform(loc=0.2, scale=0.8), \n 'colsample_bytree': sp_uniform(loc=0.4, scale=0.6)}\n\nLGBReg = lgb.LGBMRegressor(n_estimators=1500, random_state=40, silent = True, n_jobs=4,max_depth =-1)\nLGBHP_opt = RandomizedSearchCV(\n estimator=LGBReg, param_distributions=params_LGB, \n n_iter=100,\n cv=3,scoring='neg_mean_squared_error',random_state=40, refit=True, verbose=True)","execution_count":67,"outputs":[]},{"metadata":{"_uuid":"611bcee3128673455b3732df95e0ede79b1f3d64","trusted":true},"cell_type":"code","source":"LGBHP_opt.fit(train_x,train_y)","execution_count":69,"outputs":[]},{"metadata":{"_uuid":"8a3292329a18bef1059d9f06768e9c457483ed3d","trusted":true},"cell_type":"code","source":"FinalLGBParams = LGBHP_opt.best_params_\nAdditionalParams = {'objective': 'regression', 'metric':'rmse', 'verbosity':-1}\nfor i in AdditionalParams:\n FinalLGBParams[i] = AdditionalParams[i]\nFinalLGBParams","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#Save the Optimum LGB Params to increase efficiency of not having to run each time to obtain them\nSavedLGBParams = {'colsample_bytree': 0.8539721005885168,\n 'learning_rate': 0.05722266438555726,\n 'min_child_samples': 453,\n 'min_child_weight': 1,\n 'num_leaves': 35,\n 'subsample': 0.6218338710274947,\n 'objective': 'regression',\n 'metric': 'rmse',\n 'verbosity': -1}","execution_count":71,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"#### Function to Run LGBM"},{"metadata":{},"cell_type":"markdown","source":"The function defined to build the LGBM is shown below. This function utilizes early stopping to optimize the n_estimators in the model utilizing the training set and the validation set. This is an important step, as early stopping will help mitigate any overfitting issues that could arise when building this model. After the model was built, predictions were obtained using the optimum model as determined by early stopping. These predictions were returned in the function output, along with the model. "},{"metadata":{"_uuid":"9b94bfd946b13329f6cfacebac0426d32593541f","trusted":true},"cell_type":"code","source":"#Define function to use Early Stopping LightGBM model with parameters determined from RandomizedSearchCV\ndef LGBM(train_x, train_y, val_x, val_y, test_x, test_y, LGBParams):\n lgbtrain = lgb.Dataset(train_x, label=train_y)\n lgbval = lgb.Dataset(val_x, val_y)\n \n start_time = time.time()\n model = lgb.train(LGBParams, lgbtrain, 1000, valid_sets = [lgbval], early_stopping_rounds=100, verbose_eval=100)\n end_time = time.time()\n print(end_time-start_time)\n \n predictions_test= model.predict(test_x, num_iteration = model.best_iteration)\n predictions_validation = model.predict(val_x, num_iteration= model.best_iteration)\n \n return predictions_test, predictions_validation, model","execution_count":72,"outputs":[]},{"metadata":{"_uuid":"f694e8c7b3c033816639599418597c438057b65c","trusted":true},"cell_type":"code","source":"# Run LGBM\npredictions_test, predictions_validation, modellgb = LGBM(train_x, train_y, validation_x, validation_y, test_x, test_y,SavedLGBParams)","execution_count":73,"outputs":[{"output_type":"stream","text":"Training until validation scores don't improve for 100 rounds.\n[100]\tvalid_0's rmse: 1.6309\n[200]\tvalid_0's rmse: 1.62188\n[300]\tvalid_0's rmse: 1.61746\n[400]\tvalid_0's rmse: 1.61275\n[500]\tvalid_0's rmse: 1.61083\n[600]\tvalid_0's rmse: 1.60963\n[700]\tvalid_0's rmse: 1.60826\n[800]\tvalid_0's rmse: 1.60685\n[900]\tvalid_0's rmse: 1.60489\n[1000]\tvalid_0's rmse: 1.60581\nDid not meet early stopping. Best iteration is:\n[950]\tvalid_0's rmse: 1.60442\n4.241919040679932\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"#### Evaluation Metrics LGBM"},{"metadata":{},"cell_type":"markdown","source":"Relevant regression evaluation metrics were then obtained for the model for both the validation data set and testing data set. In summary, the performance of the model is slightly worse on the testing dataset for all metrics considered. This indicates that the model doesn't have any major issues generalizing, and overfitting has been mostly avoided. "},{"metadata":{"trusted":true},"cell_type":"code","source":"#Testing Set\nMSE_LGB_test = sklearn.metrics.mean_squared_error(test_y,predictions_test)\nMAE_LGB_test= sklearn.metrics.mean_absolute_error(test_y, predictions_test)\nRMSE_LGB_test = np.sqrt(MSE_LGB_test)\n#Validation Set\nMSE_LGB_val = sklearn.metrics.mean_squared_error(validation_y, predictions_validation)\nMAE_LGB_val = sklearn.metrics.mean_absolute_error(validation_y, predictions_validation)\nRMSE_LGB_val = np.sqrt(MSE_LGB_val)\nprint('Mean square Error for the LGBM testing set: {}, Mean absolute error for the LGBM testing set: {}, Root mean square error for the LGBM testing set: {}'.format(MSE_LGB_test, MAE_LGB_test, RMSE_LGB_test))\n\nprint('Mean square Error for the LGBM validation set: {}, Mean absolute error for the LGBM validation set: {}, Root mean square error for the LGBM validation set: {}'.format(MSE_LGB_val, MAE_LGB_val, RMSE_LGB_val))","execution_count":74,"outputs":[{"output_type":"stream","text":"Mean square Error for the LGBM testing set: 3.4084601891915356, Mean absolute error for the LGBM testing set: 0.4847850944275474, Root mean square error for the LGBM testing set: 1.8462015570331252\nMean square Error for the LGBM validation set: 2.5741667765353267, Mean absolute error for the LGBM validation set: 0.39738552506235636, Root mean square error for the LGBM validation set: 1.6044210097525295\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"#### Feature Importance LGBM"},{"metadata":{},"cell_type":"markdown","source":"Lastly, feature importances were obtained for the LGBM. The top 3 features as determined by the LGBM are totals.timeOnSite, geoNetwork.networkDomain, and totals.sessionQualityDim. This results seem to be reasonably intuitive after the exploratory analysis that was conducted on the dataset."},{"metadata":{"_uuid":"1843a03f6d48cac53253a4c54f3c7d4e72627055","trusted":true},"cell_type":"code","source":"#Obtain and plot figure importance for the LGBM\nfig, ax = plt.subplots(figsize=(12,18))\nlgb.plot_importance(modellgb, max_num_features=50, height=0.8, ax=ax)\nax.grid(False)\nplt.title(\"LightGBM - Feature Importance\", fontsize=15)\nplt.show()","execution_count":75,"outputs":[{"output_type":"display_data","data":{"text/plain":"
","image/png":"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\n"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"\n### XGBoost"},{"metadata":{},"cell_type":"markdown","source":"#### Hyperparameter Tuning XGBoost"},{"metadata":{},"cell_type":"markdown","source":"Similar to the LGBM, the first step completed for the XGBoost model was hyperparameter tuning with the RandomizedSearchCV. To reiterate what was previously stated, this method allows for very flexible and efficient hyperparameter tuning compared to other methods. As early stopping was again going to be used with the training and validation datasets to determine the n_estimators, the value was set to a high bound of 2500 here. The difference between the two models really showed in this step, as originally 3 cross validation folds with 100 iterations each was used here similar to the LGBM. This took way to long a time to execute. Instead, the cross validation folds was reduced to 2 with 50 iterations in each fold. This solution still took much longer to execute than the LGBM's hyperparameter tuning, at a run time of approximately 29 minutes. The tuned hyperparameters for the XGBoost model were saved in a dictionary called \"SavedXGBParams\" in the interest of efficiency, and can be seen there."},{"metadata":{"_uuid":"093062995775635960ca95c3ce99704a8ad3062f","trusted":true},"cell_type":"code","source":"# Prior to early stopping, determine best hyperparameters for most important parameters in \n#XGB model. n_estimators, one of the most important parameters, is determined by \n# early stopping, so a maximum limit is set here \nparams_XGB= { \n \"learning_rate\": sp_uniform(0.05, 0.4),\n \"colsample_bytree\": sp_uniform(loc=0.4, scale=0.6),\n 'max_depth': sp_randint(3, 40),\n \"subsample\": sp_uniform(loc=0.2, scale=0.8),\n \"gamma\": sp_uniform(0, 10),\n 'reg_alpha': [0, 1e-1, 1, 2, 5, 7, 10, 50, 100],\n \"min_child_weight\": [1e-5, 1e-3, 1e-2, 1e-1, 1, 1e1, 1e2, 1e3, 1e4],\n}\n\nXGBReg = xgb.XGBRegressor(n_estimators=2500, random_state=40, silent = True, n_jobs=4)\nXGBHP_opt = RandomizedSearchCV(\n estimator=XGBReg, param_distributions=params_XGB, \n n_iter=50,cv=2,scoring='neg_mean_squared_error',random_state=40, refit=True, verbose=True)\n","execution_count":76,"outputs":[]},{"metadata":{"_uuid":"aee3437904756cd91152e2ee6b3e5f50175e0355","trusted":true},"cell_type":"code","source":"XGBHP_opt.fit(train_x,train_y)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"FinalXGBParams = XGBHP_opt.best_params_\nFinalXGBParams","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"AdditionalParams = {'objective': 'reg:linear', 'eval_metric':'rmse', 'silent':True, 'random_state':40}\nfor i in AdditionalParams:\n FinalXGBParams[i] = AdditionalParams[i]\nFinalXGBParams","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#Save the Optimum XGB Params to increase efficiency of not having to run each time to obtain them\nSavedXGBParams = {'colsample_bytree': 0.5454951393008803,\n 'gamma': 4.834820481469037,\n 'learning_rate': 0.050158252068833425,\n 'max_depth': 36,\n 'min_child_weight': 1,\n 'reg_alpha': 100,\n 'subsample': 0.9278827742521691,\n 'objective': 'reg:linear',\n 'eval_metric': 'rmse',\n 'silent': True,\n 'random_state': 40}\nSavedXGBParams","execution_count":77,"outputs":[{"output_type":"execute_result","execution_count":77,"data":{"text/plain":"{'colsample_bytree': 0.5454951393008803,\n 'gamma': 4.834820481469037,\n 'learning_rate': 0.050158252068833425,\n 'max_depth': 36,\n 'min_child_weight': 1,\n 'reg_alpha': 100,\n 'subsample': 0.9278827742521691,\n 'objective': 'reg:linear',\n 'eval_metric': 'rmse',\n 'silent': True,\n 'random_state': 40}"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"#### Function to Run XGBoost"},{"metadata":{},"cell_type":"markdown","source":"The below function was used to run the XGBoost model on the dataset. This function utilizes early stopping with the training and validation dataset to find the optimum number of n_estimators to avoid overfitting problems. After the model is built with early stopping, predictions are obtained using the validation set and testing dataset. The function returns the final model and the predictions on both of the datasets. "},{"metadata":{"_uuid":"a53cc238fe0c8837c40fc265410c68e29d58e227","trusted":true},"cell_type":"code","source":"#Define function to run XGBoost model\ndef XGB(train_x, train_y, val_x, val_y, test_x, test_y, paramsXGB): \n xgb_train_dataMatrix = xgb.DMatrix(train_x, label = train_y)\n xgb_val_dataMatrix = xgb.DMatrix(val_x, label = val_y)\n xgb_test_dataMatrix = xgb.DMatrix(test_x, label= test_y)\n \n start_time = time.time()\n model = xgb.train(paramsXGB, xgb_train_dataMatrix, \n num_boost_round=2000, \n evals= [(xgb_train_dataMatrix, 'train'), (xgb_val_dataMatrix, 'valid')],\n early_stopping_rounds=100, \n verbose_eval=100\n ) \n end_time = time.time()\n print(end_time-start_time)\n \n y_pred_val = model.predict(xgb_val_dataMatrix, ntree_limit=model.best_ntree_limit)\n y_pred_test = model.predict(xgb_test_dataMatrix, ntree_limit=model.best_ntree_limit)\n return y_pred_test, y_pred_val, model","execution_count":78,"outputs":[]},{"metadata":{"_uuid":"41c7c54b15d1d2c082f720769c39ae14ddc11cf8","trusted":true},"cell_type":"code","source":"#Run XGBM\npredictions_testingData, predictions_validationData, model = XGB(train_x, train_y, validation_x, validation_y, test_x, test_y,SavedXGBParams)","execution_count":79,"outputs":[{"output_type":"stream","text":"[0]\ttrain-rmse:1.66014\tvalid-rmse:1.7989\nMultiple eval metrics have been passed: 'valid-rmse' will be used for early stopping.\n\nWill train until valid-rmse hasn't improved in 100 rounds.\n[100]\ttrain-rmse:1.24918\tvalid-rmse:1.61975\nStopping. Best iteration:\n[59]\ttrain-rmse:1.29716\tvalid-rmse:1.61678\n\n1.7584304809570312\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"#### Evaluation Metrics XGBoost generated on the Testing Set"},{"metadata":{},"cell_type":"markdown","source":"Relevant regression evaluation metrics were obtained on both the validation and testing dataset. Similar to the LGBM, the model performs slightly worse on the testing set compared to the validation set. This indicates that the XGBoost model does not have major issues generalizing, and overfitting has been mostly avoided. The models will be compared in the section titled \"Summary and Comparison of the Two Models\". "},{"metadata":{"trusted":true},"cell_type":"code","source":"#Testing Set\nMSE_XGB_testing = sklearn.metrics.mean_squared_error(test_y,predictions_testingData)\nMAE_XGB_testing= sklearn.metrics.mean_absolute_error(test_y, predictions_testingData)\nRMSE_XGB_testing = np.sqrt(MSE_XGB_testing)\n\n#Validation Set\nMSE_XGB_validation = sklearn.metrics.mean_squared_error(validation_y, predictions_validationData)\nMAE_XGB_validation = sklearn.metrics.mean_absolute_error(validation_y, predictions_validationData)\nRMSE_XGB_validation = np.sqrt(MSE_XGB_validation)\nprint('Mean square Error for the XGB testing set: {}, Mean absolute error for the XGB testing set: {}, Root mean square error for the XGB testing set: {}'.format(MSE_XGB_testing, MAE_XGB_testing, RMSE_XGB_testing))\n\nprint('Mean square Error for the XGB validation set: {}, Mean absolute error for the XGB validation set: {}, Root mean square error for the XGB validation set: {}'.format(MSE_XGB_validation, MAE_XGB_validation, RMSE_XGB_validation))","execution_count":80,"outputs":[{"output_type":"stream","text":"Mean square Error for the XGB testing set: 3.650399237216839, Mean absolute error for the XGB testing set: 0.35364854115773003, Root mean square error for the XGB testing set: 1.910601799752329\nMean square Error for the XGB validation set: 2.6139863567086423, Mean absolute error for the XGB validation set: 0.340946571971888, Root mean square error for the XGB validation set: 1.6167827178407872\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"#### Feature Importance XGBoost"},{"metadata":{},"cell_type":"markdown","source":"The feature importances as reported by the XGBoost model are shown in the figure below. The XGBoost reports that total.sessionQualityDim, totals.hits, and totals.timeOnSite are the three most important features for this prediction. These results also seem to be fairly intuitive after the exploratory analysis. "},{"metadata":{"_uuid":"8c1e918fd5159eddb286b337e0758219b12f9a45","trusted":true},"cell_type":"code","source":"#Obtain and plot figure importance for the LGBM\nfig, ax = plt.subplots(figsize=(12,18))\nxgb.plot_importance(model, max_num_features=50, height=0.8, ax=ax)\nax.grid(False)\nplt.title(\"XGB - Feature Importance\", fontsize=15)\nplt.show()","execution_count":81,"outputs":[{"output_type":"display_data","data":{"text/plain":"
","image/png":"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\n"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"\n### Summary and Comparison of the Two Models"},{"metadata":{},"cell_type":"markdown","source":"Firstly, the LightGBM showcased a much faster training time than the XGBoost model. This is best exemplified through the hyperparameter tuning that took place, as the LightGBM model was able to fit 3 cross validation folds with 100 possible combinations in each fold in approximately 8 minutes, while the XGBoost model fit 2 cross validation folds with 50 possible combinations in each fold in approximately 29 minutes. Next, the LightGBM model had slightly better accuracy on MSE and RMSE on both the validation set and test set, while performing slightly worse in terms of MAE on both datasets. Overall, it is clear the LightGBM model is the superior model on this dataset when considering the evaluation metrics with the time taken to build and run the models. It is also important to note that neither the lightGBM and XGBoost models seemed to generalize significantly better than the other, as both of the models prediction accuracy decreased around the same values from the validation set to the testing set. An interesting point of comparison between the models is that during early stopping, the LGBM model produced an optimum value of 950 estimators, while the XGBoost model produced an optimum value of 59 estimators.\n\nIn terms of feature importance, the models seem to have some agreement on the top 5 most important features for this regression task, as totals.sessionQualityDim, totals.timeOnSite, and geoNetwork.networkDomain recieved scores in the top 5 for both models feature importances. This consensus indicates these three features are crucial attributes in determining revenue per customer. "}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.7.1"}},"nbformat":4,"nbformat_minor":1}