{
  "id": 541533,
  "title": "About Public test data ",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/541533",
  "author_name": "",
  "post_date": "2024-10-20T03:40:18.785970Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>is 'To help you author robust submissions, during the final weeks of the model training phase we will be extending the public test set to include data closer to the submission deadline. Predictions on this extended set will not be scored' means in latest of days of the competitions the hosters will provide the competitors with new updated test data from their productions system?</p>",
  "messages": [
    {
      "id": "3022890",
      "postDate": "10/20/2024 03:40:18",
      "content": "<p>is 'To help you author robust submissions, during the final weeks of the model training phase we will be extending the public test set to include data closer to the submission deadline. Predictions on this extended set will not be scored' means in latest of days of the competitions the hosters will provide the competitors with new updated test data from their productions system?</p>",
      "rawMarkdown": "is 'To help you author robust submissions, during the final weeks of the model training phase we will be extending the public test set to include data closer to the submission deadline. Predictions on this extended set will not be scored' means in latest of days of the competitions the hosters will provide the competitors with new updated test data from their productions system?",
      "votes": null
    },
    {
      "id": "3023589",
      "postDate": "10/20/2024 18:11:31",
      "content": "<p><a href=\"https://www.kaggle.com/saidkoussi\" target=\"_blank\">@saidkoussi</a> To summarize the basics, I believe that the current training data is followed by six months of public data and six months of private data. Since this is a time-series problem, using the public data, which is closer to the private data, for training would enable more accurate modeling for the private data. Therefore, I think they will release a portion of the public data.</p>",
      "rawMarkdown": "saidkoussi To summarize the basics, I believe that the current training data is followed by six months of public data and six months of private data. Since this is a time-series problem, using the public data, which is closer to the private data, for training would enable more accurate modeling for the private data. Therefore, I think they will release a portion of the public data.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3023589,
      "author_name": "chumajin",
      "author_url": "",
      "post_date": "10/20/2024 18:11:31",
      "content": "<p><a href=\"https://www.kaggle.com/saidkoussi\" target=\"_blank\">@saidkoussi</a> To summarize the basics, I believe that the current training data is followed by six months of public data and six months of private data. Since this is a time-series problem, using the public data, which is closer to the private data, for training would enable more accurate modeling for the private data. Therefore, I think they will release a portion of the public data.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3022890": "is 'To help you author robust submissions, during the final weeks of the model training phase we will be extending the public test set to include data closer to the submission deadline. Predictions on this extended set will not be scored' means in latest of days of the competitions the hosters will provide the competitors with new updated test data from their productions system?",
    "3023589": "saidkoussi To summarize the basics, I believe that the current training data is followed by six months of public data and six months of private data. Since this is a time-series problem, using the public data, which is closer to the private data, for training would enable more accurate modeling for the private data. Therefore, I think they will release a portion of the public data."
  },
  "source": "meta"
}