{
  "id": 542309,
  "title": "Time column in train and test interpretation",
  "url": "/competitions/brist1d/discussion/542309",
  "author_name": "",
  "post_date": "2024-10-24T05:49:29.498662700Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>In the train data, it is observed that for each p_num the time column increases in 15 minutes interval and there is repetition of the time over the rows<br>\n(ie. for p_num p01 it started with 6:10:00 for p01_0 and increases over the rows then back to 06:05:00 for p01_95). <br>\nIs the time in train recorded over a few days time period?</p>\n<p>In the test data, the time column is random, with no chronology unlike train. Would it make sense to order it in increasing order of time?</p>",
  "messages": [
    {
      "id": "3026689",
      "postDate": "10/24/2024 05:49:29",
      "content": "<p>In the train data, it is observed that for each p_num the time column increases in 15 minutes interval and there is repetition of the time over the rows<br>\n(ie. for p_num p01 it started with 6:10:00 for p01_0 and increases over the rows then back to 06:05:00 for p01_95). <br>\nIs the time in train recorded over a few days time period?</p>\n<p>In the test data, the time column is random, with no chronology unlike train. Would it make sense to order it in increasing order of time?</p>",
      "rawMarkdown": "In the train data, it is observed that for each p_num the time column increases in 15 minutes interval and there is repetition of the time over the rows\n(ie. for p_num p01 it started with 6:10:00 for p01_0 and increases over the rows then back to 06:05:00 for p01_95). \nIs the time in train recorded over a few days time period?\n\nIn the test data, the time column is random, with no chronology unlike train. Would it make sense to order it in increasing order of time?",
      "votes": null
    },
    {
      "id": "3026857",
      "postDate": "10/24/2024 09:39:25",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/nicolechong\" target=\"_blank\">@nicolechong</a>,</p>\n<p>The train data covers a 3-month period and is in chronological order with overlap.</p>\n<p>The test data covers a separate 3-month period but is not in chronological order and does not overlap (This is to avoid data leakage so you can't use future blood glucose readings in the prediction).</p>\n<p>The sampling over the test period and randomisation has been performed to make it virtually impossible to reorder the test samples.</p>\n<p>Let me know if you want any further clarification.</p>",
      "rawMarkdown": "Hi @nicolechong,\n\nThe train data covers a 3-month period and is in chronological order with overlap.\n\nThe test data covers a separate 3-month period but is not in chronological order and does not overlap (This is to avoid data leakage so you can't use future blood glucose readings in the prediction).\n\nThe sampling over the test period and randomisation has been performed to make it virtually impossible to reorder the test samples.\n\nLet me know if you want any further clarification.",
      "votes": null
    },
    {
      "id": "3026964",
      "postDate": "10/24/2024 11:50:07",
      "content": "<p>thanks for the clarification</p>",
      "rawMarkdown": "thanks for the clarification",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3026857,
      "author_name": "samajames",
      "author_url": "",
      "post_date": "10/24/2024 09:39:25",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/nicolechong\" target=\"_blank\">@nicolechong</a>,</p>\n<p>The train data covers a 3-month period and is in chronological order with overlap.</p>\n<p>The test data covers a separate 3-month period but is not in chronological order and does not overlap (This is to avoid data leakage so you can't use future blood glucose readings in the prediction).</p>\n<p>The sampling over the test period and randomisation has been performed to make it virtually impossible to reorder the test samples.</p>\n<p>Let me know if you want any further clarification.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3026964,
          "author_name": "nicolechong",
          "author_url": "",
          "post_date": "10/24/2024 11:50:07",
          "content": "<p>thanks for the clarification</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3026689": "In the train data, it is observed that for each p_num the time column increases in 15 minutes interval and there is repetition of the time over the rows\n(ie. for p_num p01 it started with 6:10:00 for p01_0 and increases over the rows then back to 06:05:00 for p01_95). \nIs the time in train recorded over a few days time period?\n\nIn the test data, the time column is random, with no chronology unlike train. Would it make sense to order it in increasing order of time?",
    "3026857": "Hi @nicolechong,\n\nThe train data covers a 3-month period and is in chronological order with overlap.\n\nThe test data covers a separate 3-month period but is not in chronological order and does not overlap (This is to avoid data leakage so you can't use future blood glucose readings in the prediction).\n\nThe sampling over the test period and randomisation has been performed to make it virtually impossible to reorder the test samples.\n\nLet me know if you want any further clarification.",
    "3026964": "thanks for the clarification"
  },
  "source": "meta"
}