{
  "id": 582347,
  "title": "Need clarification regarding only IMU in test data",
  "url": "/competitions/cmi-detect-behavior-with-sensor-data/discussion/582347",
  "author_name": "",
  "post_date": "2025-05-30T16:03:33.985548200Z",
  "votes": 3,
  "comment_count": 4,
  "views": 0,
  "content": "<blockquote>\n  <p>Expect approximately 3,500 sequences in the hidden test set. Half of the test data will only include IMU sensor data, while the other half will contain data from all sensors.</p>\n</blockquote>\n<p>Does this mean the whole test data will be of 3500 length, public+private or is it only about the private leaderboard?<br>\nSecodnly, will the other half not have the sensor data other than IMU at all, does this mean we need to hadnly missing columns or use two models?</p>",
  "messages": [
    {
      "id": "3213884",
      "postDate": "05/30/2025 16:03:33",
      "content": "<blockquote>\n  <p>Expect approximately 3,500 sequences in the hidden test set. Half of the test data will only include IMU sensor data, while the other half will contain data from all sensors.</p>\n</blockquote>\n<p>Does this mean the whole test data will be of 3500 length, public+private or is it only about the private leaderboard?<br>\nSecodnly, will the other half not have the sensor data other than IMU at all, does this mean we need to hadnly missing columns or use two models?</p>",
      "rawMarkdown": "> Expect approximately 3,500 sequences in the hidden test set. Half of the test data will only include IMU sensor data, while the other half will contain data from all sensors.\n\nDoes this mean the whole test data will be of 3500 length, public+private or is it only about the private leaderboard?\nSecodnly, will the other half not have the sensor data other than IMU at all, does this mean we need to hadnly missing columns or use two models?",
      "votes": null
    },
    {
      "id": "3213908",
      "postDate": "05/30/2025 16:35:02",
      "content": "<p><a href=\"https://www.kaggle.com/abhirupchoudhury073\" target=\"_blank\">@abhirupchoudhury073</a> please refer to this link. <a href=\"https://www.kaggle.com/competitions/cmi-detect-behavior-with-sensor-data/discussion/582334#3213893\" target=\"_blank\">https://www.kaggle.com/competitions/cmi-detect-behavior-with-sensor-data/discussion/582334#3213893</a></p>\n<p>But we don't know about public or private.</p>",
      "rawMarkdown": "abhirupchoudhury073 please refer to this link. https://www.kaggle.com/competitions/cmi-detect-behavior-with-sensor-data/discussion/582334#3213893\n\nBut we don't know about public or private.",
      "votes": null
    },
    {
      "id": "3213983",
      "postDate": "05/30/2025 17:57:30",
      "content": "<p><a href=\"https://www.kaggle.com/abhirupchoudhury073\" target=\"_blank\">@abhirupchoudhury073</a> , We've updated the documentation: Half of the hidden-test sequences are recorded with IMU only; the thermopile and time-of-flight columns are still present but contain null values for those sequences. We don't reveal info about public vs private.</p>",
      "rawMarkdown": "abhirupchoudhury073 , We've updated the documentation: Half of the hidden-test sequences are recorded with IMU only; the thermopile and time-of-flight columns are still present but contain null values for those sequences. We don't reveal info about public vs private.",
      "votes": null
    },
    {
      "id": "3214982",
      "postDate": "06/01/2025 11:26:45",
      "content": "<p>If you need to know, you can manually modify those columns in test set to missing values, compare the score difference before and after modification, and calculate the proportion of missing values in the public leaderboard.</p>",
      "rawMarkdown": "If you need to know, you can manually modify those columns in test set to missing values, compare the score difference before and after modification, and calculate the proportion of missing values in the public leaderboard.",
      "votes": null
    },
    {
      "id": "3217950",
      "postDate": "06/05/2025 15:20:31",
      "content": "<p><a href=\"https://www.kaggle.com/martynaplomecka\" target=\"_blank\">@martynaplomecka</a> </p>\n<p>I have two questions. I'd like to ask to be precise.</p>\n<ol>\n<li><p>For the 'IMU only' sequences, when the description says they 'contain null values,' does this mean that only some of the values in the thermopile and time-of-flight columns are null, not necessarily all of them?</p></li>\n<li><p>Conversely, for the other half of the sequences (the non-IMU-only ones), does this mean their thermopile and time-of-flight columns are complete and contain no null values at all?</p></li>\n</ol>\n<p>I just want to make sure I'm interpreting this correctly.</p>",
      "rawMarkdown": "martynaplomecka \n\nI have two questions. I'd like to ask to be precise.\n\n1. For the 'IMU only' sequences, when the description says they 'contain null values,' does this mean that only some of the values in the thermopile and time-of-flight columns are null, not necessarily all of them?\n\n2. Conversely, for the other half of the sequences (the non-IMU-only ones), does this mean their thermopile and time-of-flight columns are complete and contain no null values at all?\n\nI just want to make sure I'm interpreting this correctly.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3213908,
      "author_name": "chumajin",
      "author_url": "",
      "post_date": "05/30/2025 16:35:02",
      "content": "<p><a href=\"https://www.kaggle.com/abhirupchoudhury073\" target=\"_blank\">@abhirupchoudhury073</a> please refer to this link. <a href=\"https://www.kaggle.com/competitions/cmi-detect-behavior-with-sensor-data/discussion/582334#3213893\" target=\"_blank\">https://www.kaggle.com/competitions/cmi-detect-behavior-with-sensor-data/discussion/582334#3213893</a></p>\n<p>But we don't know about public or private.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3213983,
      "author_name": "martynaplomecka",
      "author_url": "",
      "post_date": "05/30/2025 17:57:30",
      "content": "<p><a href=\"https://www.kaggle.com/abhirupchoudhury073\" target=\"_blank\">@abhirupchoudhury073</a> , We've updated the documentation: Half of the hidden-test sequences are recorded with IMU only; the thermopile and time-of-flight columns are still present but contain null values for those sequences. We don't reveal info about public vs private.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3217950,
          "author_name": "yudaiueno",
          "author_url": "",
          "post_date": "06/05/2025 15:20:31",
          "content": "<p><a href=\"https://www.kaggle.com/martynaplomecka\" target=\"_blank\">@martynaplomecka</a> </p>\n<p>I have two questions. I'd like to ask to be precise.</p>\n<ol>\n<li><p>For the 'IMU only' sequences, when the description says they 'contain null values,' does this mean that only some of the values in the thermopile and time-of-flight columns are null, not necessarily all of them?</p></li>\n<li><p>Conversely, for the other half of the sequences (the non-IMU-only ones), does this mean their thermopile and time-of-flight columns are complete and contain no null values at all?</p></li>\n</ol>\n<p>I just want to make sure I'm interpreting this correctly.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3214982,
      "author_name": "yunsuxiaozi",
      "author_url": "",
      "post_date": "06/01/2025 11:26:45",
      "content": "<p>If you need to know, you can manually modify those columns in test set to missing values, compare the score difference before and after modification, and calculate the proportion of missing values in the public leaderboard.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3213884": "> Expect approximately 3,500 sequences in the hidden test set. Half of the test data will only include IMU sensor data, while the other half will contain data from all sensors.\n\nDoes this mean the whole test data will be of 3500 length, public+private or is it only about the private leaderboard?\nSecodnly, will the other half not have the sensor data other than IMU at all, does this mean we need to hadnly missing columns or use two models?",
    "3213908": "abhirupchoudhury073 please refer to this link. https://www.kaggle.com/competitions/cmi-detect-behavior-with-sensor-data/discussion/582334#3213893\n\nBut we don't know about public or private.",
    "3213983": "abhirupchoudhury073 , We've updated the documentation: Half of the hidden-test sequences are recorded with IMU only; the thermopile and time-of-flight columns are still present but contain null values for those sequences. We don't reveal info about public vs private.",
    "3214982": "If you need to know, you can manually modify those columns in test set to missing values, compare the score difference before and after modification, and calculate the proportion of missing values in the public leaderboard.",
    "3217950": "martynaplomecka \n\nI have two questions. I'd like to ask to be precise.\n\n1. For the 'IMU only' sequences, when the description says they 'contain null values,' does this mean that only some of the values in the thermopile and time-of-flight columns are null, not necessarily all of them?\n\n2. Conversely, for the other half of the sequences (the non-IMU-only ones), does this mean their thermopile and time-of-flight columns are complete and contain no null values at all?\n\nI just want to make sure I'm interpreting this correctly."
  },
  "source": "meta"
}