{
  "id": 556356,
  "title": "number of time_id in test data",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/556356",
  "author_name": "",
  "post_date": "2025-01-12T21:21:55.063607100Z",
  "votes": -2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>In the test data, are we supposed to predict responders for one day based on the lag.parquet? In other words, it seems that the test set provided every time is a one-day dataset, with the corresponding lags.parquet provided together. Am I right?</p>",
  "messages": [
    {
      "id": "3095045",
      "postDate": "01/12/2025 21:21:55",
      "content": "<p>In the test data, are we supposed to predict responders for one day based on the lag.parquet? In other words, it seems that the test set provided every time is a one-day dataset, with the corresponding lags.parquet provided together. Am I right?</p>",
      "rawMarkdown": "In the test data, are we supposed to predict responders for one day based on the lag.parquet? In other words, it seems that the test set provided every time is a one-day dataset, with the corresponding lags.parquet provided together. Am I right?",
      "votes": null
    },
    {
      "id": "3095101",
      "postDate": "01/13/2025 00:38:58",
      "content": "<p>You're supposed to make predictions based on the test dataframe. Lags are provided for either feature engineering or assessing your prediction accuracy for the previous day. You only get given lags for the test data at time_id == 0 for each day. The other 967 time steps each day you won't recieve lags</p>\n<p>See some test code here you can drop in your notebook and it will call your predict function in a way that simulates the competition format <a href=\"https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/555226\" target=\"_blank\">https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/555226</a></p>",
      "rawMarkdown": "You're supposed to make predictions based on the test dataframe. Lags are provided for either feature engineering or assessing your prediction accuracy for the previous day. You only get given lags for the test data at time_id == 0 for each day. The other 967 time steps each day you won't recieve lags\n\n\nSee some test code here you can drop in your notebook and it will call your predict function in a way that simulates the competition format https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/555226",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3095101,
      "author_name": "michaeltimbs",
      "author_url": "",
      "post_date": "01/13/2025 00:38:58",
      "content": "<p>You're supposed to make predictions based on the test dataframe. Lags are provided for either feature engineering or assessing your prediction accuracy for the previous day. You only get given lags for the test data at time_id == 0 for each day. The other 967 time steps each day you won't recieve lags</p>\n<p>See some test code here you can drop in your notebook and it will call your predict function in a way that simulates the competition format <a href=\"https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/555226\" target=\"_blank\">https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/555226</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3095045": "In the test data, are we supposed to predict responders for one day based on the lag.parquet? In other words, it seems that the test set provided every time is a one-day dataset, with the corresponding lags.parquet provided together. Am I right?",
    "3095101": "You're supposed to make predictions based on the test dataframe. Lags are provided for either feature engineering or assessing your prediction accuracy for the previous day. You only get given lags for the test data at time_id == 0 for each day. The other 967 time steps each day you won't recieve lags\n\n\nSee some test code here you can drop in your notebook and it will call your predict function in a way that simulates the competition format https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/555226"
  },
  "source": "meta"
}