Many organizations prospect for loyal supporters and donors by sending direct mail appeals. This is an effective way to build a large base, but can be very expensive and have a low efficiency. Eliminating likely non-donors is the key to running an efficient\r\n Prospecting program and ultimately to pursuing the mission of the organization. Help us help these organizations to target the best prospective donors and fund organizational goals!
\r\nPlease note: This competition has rules that we have not used previously restricting what kinds of models are acceptable. Please see the rules page for more information.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "rules", "content": "\r\n\t\r\nYou cannot sign up to Kaggle from multiple accounts and therefore you cannot submit from multiple accounts.
\r\n\r\n Privately sharing code or data outside of teams is not permitted.\r\n It's okay to share code if made available to all participants on the forums.\r\n
\r\n\r\n Team mergers are allowed and can be performed by the team leader. In order to merge, the combined team must have a total submission count less than or equal to the maximum allowed as of the merge date. The maximum allowed is the number of submissions per day multiplied by the number of days the competition has been running.
\r\n
There is no maximum team size.
\r\nYou may submit a maximum of 2 entries per day.
\r\nYou may select up to 5 final submissions for judging.
\r\nStart Date: 7/18/2012 9:39 PM UTC
\r\nMerger Deadline: None
\r\nEntry Deadline: None
\r\nEnd Date: 9/18/2012 11:59 PM UTC
\r\n\t\r\nPlease post any questions you have about these rules in the forum.
\r\nThe data come from three different datasets (given by the databaseid variable in each dataset). For privacy reasons, individual giving history from DB1 and DB3 may not be used together. Using this information in the aggregate is fine; for\r\n example, a model may notice based on all of the combined information that prospects in particular zip codes are good/bad. But using information in DB1 to identify an individual prospect as good/bad and then using that to make predictions about a record in\r\n DB3 is not permitted.
\r\nThis restriction applies to using information in DB1 for predicitons about DB3 and vice versa, but does not apply to DB2 in either direction.
\r\nOutside data may be used only if publicly available. Descriptions of the data and where to find it must be posted in the forum, and the host may approve or disapprove the use of the data.
\r\nFinal solution must be able to be implemented using open source statistical tools including SQL, Python, R, pmml, etc. Matlab will also be accepted due to the relatively low license cost.
\r\nContestants must not try to deanonymize any aspect of the data, including determining the identities of donors or the identities of soliciting organizations.
\r\nThe data provided may only be used for the purposes of preparing entries to this competition.
\r\nMore detail on the above is included in the FAQ, specifically questions 4-6.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Background", "content": "Thousands of non-profit organizations and political campaigns originate their loyal donors through low-dollar fundraising appeals in the mail, over the phone, and online. These channels tend to be low efficiency, but are essential to building a base of loyal\r\n donors who will eventually be migrated up the giving ladder and into the major gift streams.
\r\nThe better the organization is at targeting the right prospects in their initial acquisition programs, the better they will ultimately be able to fund the pursuit of the Mission.
\r\nOur goal is to help these organizations build their donor bases in the most efficient way possible. We aim to help them target appeals to the most receptive donors. This competition is specifically addressing direct mail campaigns - making them as efficient\r\n as possible. Future competitions will address the challenge of migrating those donors into new categories.
\r\nA successful model will have an incredible impact on the ability of these organizations to continute to operate, and to be positioned to effect important change around the world by providing the financial resources to pursue their Missions.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Evaluation", "content": "Note: The maximum possible public leaderboard score is: 0.87190
\r\nThe response variable is "Amount2" (donation amount raised to the 1.15 power), and the metric used is "AverageAmongTopP" (with P=75%). This response variable is not included in the training dataset,\r\n but you can construct it (as in the sample code).
\r\nThe Solution Data is composed of many individual mailings. We are not trying to achieve a result that recommends "Mail all of package 1 and none of package 2." Each mailing will be mailed - the goal of this contest is to mail the best prospects for EACH\r\n mailing. Therefore, prospects in each mailing will be rank-ordered based on the model output. The top 75% of each mailing, according to the model results, will be categorized as "good" prospects. The bottom 25% will be categorized as "bad" prospects. Category\r\n variables: "ListID," "Package," and "Agency" may be used as variables in the scoring algorithm, but using them to identify superior overall mailings (should be mailed 100%) and inferior mailings (should not be mailed) will not achieve the goal of maximizing\r\n model performance for each mailing.
\r\nThe Training Data will be from Nov 2011 and earlier. The Solution Data will be from Dec 2011 and into 2012.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Prizes", "content": "First prize:$6000
\r\nSecond prize:$3000
\r\nThird prize: $1000
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Terms", "content": "Participants in the competition will grant a license for use of their model to the Host. This license will be exclusive within the domain of fund-raising, but will be non-exclusive in any other domains.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "data-description", "content": "You will predict "Amount2", which is a transformation of the donation amount (donation amount raised to the 1.15 power).\r\n
\r\nTraining data: kaggle_training_dataset_formatted2
\r\nTesting data: test.
\r\nPlease see "Kaggle FAQ" (downloadable with the data files) for other questions.
\r\nNote: Category variables ("ListID," "Package," and "Agency") may be used as variables in the scoring algorithm, but using them to identify superior overall mailings (should be mailed 100%) and inferior mailings (should not be mailed) will\r\n not achieve the goal of maximizing model performance for each mailing. This is because we will take the top 75% of prospects in *each mailing* when evaluating performance.
\r\n\r\n
TABLE OVERVIEW
\r\nKaggle_training_dataset_formatted2: Full mail history for the 11 months leading up to the Solution data.
\r\nKaggle_donation_dataset_formatted2: Entire donation history pre-Training dataset for all organizations in Agency 1, 2 and 3
\r\nKaggle_mail_dataset_formatted1a: Part of mail history before Training_dataset for Agency 1
\r\nKaggle_mail_dataset_formatted1b: Second part of Agency 1 Training_dataset
\r\nKaggle_mail_dataset_formatted2: mail history before Training_dataset for Agency 2
\r\nKaggle_mail_dataset_formatted3: mail history before Training_dataset for Agency 3
\r\nDemo_per_formatted: Demographic information by 9-digit zip code
\r\nZip_perf: summary of historical mail performance by 5-digit zip
\r\ntraining_sample: This is a random 10% sample of the training data, provided for your convenience. The sample submission is based on this dataset rather than the full training dataset.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Winners", "content": "The winners are:
\r\n