{
  "id": 393809,
  "title": "Field Descriptions",
  "url": "/competitions/tlvmc-parkinsons-freezing-gait-prediction/discussion/393809",
  "author_name": "",
  "post_date": "2023-03-10T20:57:11.405540900Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I'm a little confused by the data description. How should Event = 0 (or StartHesitation=0, Turn=0, Walking=0) while Valid = true and Task = True be interpreted?</p>",
  "messages": [
    {
      "id": "2176681",
      "postDate": "03/10/2023 20:57:11",
      "content": "<p>I'm a little confused by the data description. How should Event = 0 (or StartHesitation=0, Turn=0, Walking=0) while Valid = true and Task = True be interpreted?</p>",
      "rawMarkdown": "I'm a little confused by the data description. How should Event = 0 (or StartHesitation=0, Turn=0, Walking=0) while Valid = true and Task = True be interpreted?",
      "votes": null
    },
    {
      "id": "2176702",
      "postDate": "03/10/2023 21:11:54",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/imlovingit\" target=\"_blank\">@imlovingit</a>,</p>\n<p>In <code>StartHesitation</code>, <code>Turn</code>, and <code>Walking</code> are the class labels for each time step. The goal of the competition is to predict these class labels for the test set series.</p>\n<p><code>Valid</code> and <code>Task</code> are metadata associated with series in the DeFOG dataset. The short version is that both have to be <code>true</code> in order for the class labels at that step to be meaningful. When constructing training or evaluation data, you might wish to drop any part of a series where either is false. These columns aren't present in the test set, but the metric is aware of them and will ignore predictions where either is actually false.</p>",
      "rawMarkdown": "Hi @imlovingit,\n\nIn `StartHesitation`, `Turn`, and `Walking` are the class labels for each time step. The goal of the competition is to predict these class labels for the test set series.\n\n`Valid` and `Task` are metadata associated with series in the DeFOG dataset. The short version is that both have to be `true` in order for the class labels at that step to be meaningful. When constructing training or evaluation data, you might wish to drop any part of a series where either is false. These columns aren't present in the test set, but the metric is aware of them and will ignore predictions where either is actually false.",
      "votes": null
    },
    {
      "id": "2176780",
      "postDate": "03/10/2023 22:29:29",
      "content": "<p>Valid = true and Task = true means that given datapoint was labelled and is valid with high confidence. The fact that Event = 0 means that it is of negative class (none of StartHesitation, Turn, Walking happened)</p>",
      "rawMarkdown": "Valid = true and Task = true means that given datapoint was labelled and is valid with high confidence. The fact that Event = 0 means that it is of negative class (none of StartHesitation, Turn, Walking happened)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2176702,
      "author_name": "ryanholbrook",
      "author_url": "",
      "post_date": "03/10/2023 21:11:54",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/imlovingit\" target=\"_blank\">@imlovingit</a>,</p>\n<p>In <code>StartHesitation</code>, <code>Turn</code>, and <code>Walking</code> are the class labels for each time step. The goal of the competition is to predict these class labels for the test set series.</p>\n<p><code>Valid</code> and <code>Task</code> are metadata associated with series in the DeFOG dataset. The short version is that both have to be <code>true</code> in order for the class labels at that step to be meaningful. When constructing training or evaluation data, you might wish to drop any part of a series where either is false. These columns aren't present in the test set, but the metric is aware of them and will ignore predictions where either is actually false.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2176780,
      "author_name": "kretes",
      "author_url": "",
      "post_date": "03/10/2023 22:29:29",
      "content": "<p>Valid = true and Task = true means that given datapoint was labelled and is valid with high confidence. The fact that Event = 0 means that it is of negative class (none of StartHesitation, Turn, Walking happened)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2176681": "I'm a little confused by the data description. How should Event = 0 (or StartHesitation=0, Turn=0, Walking=0) while Valid = true and Task = True be interpreted?",
    "2176702": "Hi @imlovingit,\n\nIn `StartHesitation`, `Turn`, and `Walking` are the class labels for each time step. The goal of the competition is to predict these class labels for the test set series.\n\n`Valid` and `Task` are metadata associated with series in the DeFOG dataset. The short version is that both have to be `true` in order for the class labels at that step to be meaningful. When constructing training or evaluation data, you might wish to drop any part of a series where either is false. These columns aren't present in the test set, but the metric is aware of them and will ignore predictions where either is actually false.",
    "2176780": "Valid = true and Task = true means that given datapoint was labelled and is valid with high confidence. The fact that Event = 0 means that it is of negative class (none of StartHesitation, Turn, Walking happened)"
  },
  "source": "meta"
}