{
  "fetched_at": "2026-10-19T22:01:24Z",
  "source_url": "https://www.kaggle.com/competitions/estmv2020/data",
  "pages": [
    {
      "name": "rules",
      "content": "This is a page where you can include rules that participants must accept before joining. You may wish to include rules like:\n\n - Don't cheat!\n - Apply yourself!\n - Have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "data-description",
      "content": "This dataset summarizes a heterogeneous set of features about articles published by Mashable in a period of two years. The goal is to predict if the article is among the most popular ones based on sharing in social networks (coded by the variable \"is_popular\").\n\nThe articles were published by Mashable (www.mashable.com) and their content as the rights to reproduce it belongs to them. Hence, this dataset does not share the original content but some statistics associated with it. Acquisition date: January 8, 2015.\n\n**File descriptions**\n\n- train.csv - the training set\n\n- test.csv - the test set\n\n- sample_submission.csv - a sample submission file in the correct format\n\n**Data fields**\n\nn_tokens_title: Number of words in the title\n\nn_tokens_content: Number of words in the content\n\nn_unique_tokens: Rate of unique words in the content\n\nn_non_stop_words: Rate of non-stop words in the content\n\nn_non_stop_unique_tokens: Rate of unique non-stop words in the content\n\nnum_hrefs: Number of links\n\nnum_self_hrefs: Number of links to other articles published by Mashable\n\nnum_imgs: Number of images\n\nnum_videos: Number of videos\n\naverage_token_length: Average length of the words in the content\n\nnum_keywords: Number of keywords in the metadata\n\ndata_channel_is_lifestyle: Is data channel 'Lifestyle'?\n\ndata_channel_is_entertainment: Is data channel 'Entertainment'?\n\ndata_channel_is_bus: Is data channel 'Business'?\n\ndata_channel_is_socmed: Is data channel 'Social Media'?\n\ndata_channel_is_tech: Is data channel 'Tech'?\n\ndata_channel_is_world: Is data channel 'World'?\n\nkw_min_min: Worst keyword (min. shares)\n\nkw_max_min: Worst keyword (max. shares)\n\nkw_avg_min: Worst keyword (avg. shares)\n\nkw_min_max: Best keyword (min. shares)\n\nkw_max_max: Best keyword (max. shares)\n\nkw_avg_max: Best keyword (avg. shares)\n\nkw_min_avg: Avg. keyword (min. shares)\n\nkw_max_avg: Avg. keyword (max. shares)\n\nkw_avg_avg: Avg. keyword (avg. shares)\n\nself_reference_min_shares: Min. shares of referenced articles in Mashable\n\nself_reference_max_shares: Max. shares of referenced articles in Mashable\n\nself_reference_avg_sharess: Avg. shares of referenced articles in Mashable\n\nweekday_is_monday: Was the article published on a Monday?\n\nweekday_is_tuesday: Was the article published on a Tuesday?\n\nweekday_is_wednesday: Was the article published on a Wednesday?\n\nweekday_is_thursday: Was the article published on a Thursday?\n\nweekday_is_friday: Was the article published on a Friday?\n\nweekday_is_saturday: Was the article published on a Saturday?\n\nweekday_is_sunday: Was the article published on a Sunday?\n\nis_weekend: Was the article published on the weekend?\n\nLDA_00: Closeness to LDA topic 0\n\nLDA_01: Closeness to LDA topic 1\n\nLDA_02: Closeness to LDA topic 2\n\nLDA_03: Closeness to LDA topic 3\n\nLDA_04: Closeness to LDA topic 4\n\nglobal_subjectivity: Text subjectivity\n\nglobal_sentiment_polarity: Text sentiment polarity\n\nglobal_rate_positive_words: Rate of positive words in the content\n\nglobal_rate_negative_words: Rate of negative words in the content\n\nrate_positive_words: Rate of positive words among non-neutral tokens\n\nrate_negative_words: Rate of negative words among non-neutral tokens\n\navg_positive_polarity: Avg. polarity of positive words\n\nmin_positive_polarity: Min. polarity of positive words\n\nmax_positive_polarity: Max. polarity of positive words\n\navg_negative_polarity: Avg. polarity of negative words\n\nmin_negative_polarity: Min. polarity of negative words\n\nmax_negative_polarity: Max. polarity of negative words\n\ntitle_subjectivity: Title subjectivity\n\ntitle_sentiment_polarity: Title polarity\n\nabs_title_subjectivity: Absolute subjectivity level\n\nabs_title_sentiment_polarity: Absolute polarity level\n\nshares: Number of shares (target)",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "This is an inclass competition for students at FEUP - MIEGI taking Estatística Mulivariada. \nDetailed assignment instructions are given in the [course Moodle page](https://moodle.up.pt/course/view.php?id=1948).\n\nIn this competition your task is to predict the articles'  number of shares in social media. The data comes from website mashable.com from beginning of 2015. The dataset used in the competition can be found at the [UCI repository](https://archive.ics.uci.edu/ml/datasets/Online+News+Popularity#). Please note, the articles number of shares have been randomly changed to prevent cheating.\n\nThere is a public leaderboard which is computed on about 30% of the test data and a private one that is computed on the remaining observations. You can use the public leaderboard to evaluate your current standing till the deadline date. The final rankings will be based on the private leaderboard. Bear in mind that by submitting many times to the public leaderboard and making decisions upon these results you might **overfit your model**.\n\n**Acknowledgments**\n\nThe dataset used in the competition can be found at the UCI repository. We thank Kelwin Fernandes, Pedro Vinagre, Paulo Cortez and Pedro Sernadela for making it publicly available. Check also their publication below.\nK. Fernandes, P. Vinagre and P. Cortez. A Proactive Intelligent Decision Support System for Predicting the Popularity of Online News. Proceedings of the 17th EPIA 2015 - Portuguese Conference on Artificial Intelligence, September, Coimbra, Portugal.\n\n\nCompetition adapted from  \"Data Science and Machine Learning 3: Tools\" course, part of the [MSc in Business Analytics](https://economics.ceu.edu/program/master-science-business-analytics) program at CEU in the Winter semester of 2018/19 (https://www.kaggle.com/c/online-news-popularity/)",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "**Steps**\n\nDevelop a regression model that fits the data in \"train.csv,\" a dataset of 5000 news (Id numbers 1-5000), by using \"shares\" as the Y variable and your choice of X variables.\n\nNext you will use the fitted regression model to predict the number of shares of the 1000 news in the \"validation.csv\" dataset. For each Id in the validation data set (numbers 5001-6000), you must predict the value of the \"shares\" variable. Each group can submit limited set of predictions up to 3 times per day.\n\n**Goal**\n\nYour team's goal is to obtain the best predictions of the news' \"shares\" in \"validation.csv\" dataset. In this competition, best is the submission with the lowest Root-Mean-Squared-Error (RMSE). This is a measure of the difference between your predictions and the true (but unknown to you) number of shares. \n\n\n**Submission Format**\n\nThe files you submit on Kaggle must contain a header row and have two columns in the following format:\n\n| Id | Prediction |\n| --- | --- |\n| 5001 | 101 |\n| 5002 | 104.6 |\n| 5003 | 1011 |\n| ... | ... |\n| 6000 | 2 |\n\nSubmissions MUST be .csv files. The easiest way to be sure you are submitting an acceptable set of predictions is to:\n\n1. download the example submission file (sample-upload.csv) from the Kaggle Data page;\n2. open that file with Excel and rename it so you can keep track of your predictions on individual worksheets (e.g.; prediction1, prediction2, etc.);\n3. make your \"shares\" predictions;\n4. copy the 1000 predictions into your Excel file created in #2 above so you have a record of what you submitted;\n5. create a CSV file and submit it.\n\n**Leaderboard**\n\nYour progress will be shown on the leaderboard, which shows your \"public\" score based on a randomly selected 30% of the test data. There is also a \"private\" leaderboard score based on the the other 70% of the test data. This score will not be revealed until the due date has passed.\n\nThere are several benchmarks given on the leaderboard to show you what poor, weak and excellent scores look like:\n\n*poor score*: Average;\n*weak score*: 6Vars (multiple regression with six quantitative variables), 6+1Vars (multiple regression with six quantitative variables and one qualitative variable);\n*excellent scores*: XGBoost, Lasso, GBM (state of the art algorithms without any fine tuning).\nYou should aim to be clearly above weak scores and close to the excellent ones.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}