{
  "fetched_at": "2026-10-18T13:44:40Z",
  "source_url": "https://www.kaggle.com/competitions/bgu-recommendation-systems-click-through-rate/data",
  "pages": [
    {
      "name": "data-description",
      "content": "#Taboola Notice\nAll Datasets and other materials provided by Taboola.com Ltd., (the “Content”) are provided only for educational and informational purposes. The Content confidential, privileged and is the property of Taboola.com Ltd, its subsidiaries, and affiliates (collectively, “Taboola”). You may not copy, reproduce, distribute, publish, display, perform, modify, create derivative works, transmit, or in any way exploit any such Content, nor may you distribute any part of this Content over any network, including a local area network, sell or offer it for sale, or use such Content to construct any kind of database other than as explicitly permitted by Taboola. Copying, storing or disseminating any Content except as provided above, intentionally or unintentionally, is expressly prohibited without prior written permission from Taboola. For permission to use the Content, please contact Legal@taboola.com. The information contained in the Content is provided on an “as is” basis with no guarantees of completeness, accuracy or usefulness and Taboola shall not be liable for any actions taken in reliance thereon.\n\n\nThe train data contains supervised data for CTR prediction of specific time, user and item.\n\n## Your submission should follow the format of the sample submission file: Id , Predicted.\nId - you do not suppose to change/filter/sort.\nPredicted - Predict the click probability [0-1].\n###Available Features\n#### Target item features\ntarget_id_hash - a unique identifier of a specific item\nsyndicator_id_hash - sponsored items are provided by syndicators (also called advertisers), and a syndicator_id is a\nunique identifier of the syndicator of the current item.\ncampaign_id_hash - syndicators promote campaign which may include a set of items. So each item has a single campaign, and the campaign_id_hash is a unique identifier of the item’s campaign.\nempiric_calibrated_recs - the number of views of the current item calibrated\nempiric_clicks - the number of clicks on the target item\ntarget_item_taxonomy - 3rd party hierarchical taxonomy of the target\nplacement_id_hash - Identifies the location in the publisher site in which the ad has been viewed\npage_view_start_time\n#### User features\nuser_id_hash - a unique identifier of a specific user\nuser_target_recs - the number of times the user saw this target\nuser_recs - total number of items presented to the user so far\nuser_clicks - total number of items clicked by the user so far\n#### Context features\npublisher_id_hash - a unique identifier of the source’s publisher\nsource_id_hash -a unique identifier of the current source\nsource_item_type - the type of the source item - Text / Photo Gallery / Video etc.\nbrowser_platform - Desktop / Mobile / Tablet etc.\nos_family - Windows / Linux / IOs / Android etc.\ncountry_code\nregion\nday_of_week\ntime_of_day\ngmt_offset\n\n# Note:\nAt the test file, there is a column named 'Id'. This column represents the original Id of the record within the original dataset (not only cold users).\nYou may use it if you want to.\nHowever, when you are submitting your prediction use the index of the test file as the Id.\n\n# In other words:\npred = clf.predict_proba(test_df)[:,1]\npred_df = pd.DataFrame(pred)\npred_df.reset_index(inplace=True)\npred_df.columns = ['Id','Predicted']\npred_df.to_csv('my_submission.csv',index=False)",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "In this challenge, we provide a data set from our large-scale recommendation system, which serves billions of recommendations per day. Each row represents an event of recommendation and contains information about the user, the context, and the recommended item. The rows are labeled as clicked/ not clicked. Your task is to build a model that predicts if a user will click on an item.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "The evaluation metric for this competition is [Area Under the Curve (AUC)](https://towardsdatascience.com/understanding-auc-roc-curve-68b2303cc9c5). AUC - ROC curve is a performance measurement for the classification problem at various threshold settings. ROC is a probability curve and AUC represents the degree or measure of separability. It tells how much the model is capable of distinguishing between classes. Higher the AUC, the better the model is at predicting 0s as 0s and 1s as 1s.\n\n\n## Submission Format\nFor every index in the dataset, submission files should contain two columns: Id and Predicted probability between [0,1]. The file should contain a header and have the following format: Id, Predicted 1,0.3 2,0.8 3,0.4 4,0.7 etc.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "One account per participant You cannot sign up to Kaggle from multiple accounts and therefore you cannot submit from multiple accounts.\nNo private sharing outside teams Privately sharing code or data outside of teams is not permitted. It's okay to share code if made available to all participants on the forums.\n\nTeam Mergers - Team mergers are not allowed in this competition.\nTeam Limits - Maximum team size 2 except miloim which can be 3.\n\nSubmission Limits - You may submit a maximum of 5 entries per day.\n\nDon't cheat!\nApply yourself!\nHave fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}