{
  "id": 408072,
  "title": "Is this a multiclass problem?",
  "url": "/competitions/tlvmc-parkinsons-freezing-gait-prediction/discussion/408072",
  "author_name": "",
  "post_date": "2023-05-09T10:15:11.178283600Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Am I right that this competition is a multiclass problem, not multilabel one? Meaning that only one class from ['StartHesitation', 'Turn' , 'Walking'] may be assigned to a specific record.</p>",
  "messages": [
    {
      "id": "2251418",
      "postDate": "05/09/2023 10:15:11",
      "content": "<p>Am I right that this competition is a multiclass problem, not multilabel one? Meaning that only one class from ['StartHesitation', 'Turn' , 'Walking'] may be assigned to a specific record.</p>",
      "rawMarkdown": "Am I right that this competition is a multiclass problem, not multilabel one? Meaning that only one class from ['StartHesitation', 'Turn' , 'Walking'] may be assigned to a specific record.",
      "votes": null
    },
    {
      "id": "2251552",
      "postDate": "05/09/2023 13:20:42",
      "content": "<p>You have three labels here, namely ['StartHesitation', 'Turn' , 'Walking'], and for each label you have several classes (I saw 2 classes for each, 0 and 1, until now, in the training set). For me we have multi class and multi label.</p>\n<p>In this case you have several possibilities : </p>\n<ol>\n<li>Three models : One model for each label (use any classifier)</li>\n<li>Build a neural network and set the output layer with dimension 3.</li>\n</ol>",
      "rawMarkdown": "You have three labels here, namely ['StartHesitation', 'Turn' , 'Walking'], and for each label you have several classes (I saw 2 classes for each, 0 and 1, until now, in the training set). For me we have multi class and multi label.\n\nIn this case you have several possibilities : \n\n1. Three models : One model for each label (use any classifier)\n2. Build a neural network and set the output layer with dimension 3.",
      "votes": null
    },
    {
      "id": "2251608",
      "postDate": "05/09/2023 14:08:50",
      "content": "<p>I use 4 classes, \"Nothing\" included- to predict there is no event right now.</p>",
      "rawMarkdown": "I use 4 classes, \"Nothing\" included- to predict there is no event right now.",
      "votes": null
    },
    {
      "id": "2251728",
      "postDate": "05/09/2023 15:31:41",
      "content": "<p>I also use 4 classes in my model.</p>",
      "rawMarkdown": "I also use 4 classes in my model.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2251552,
      "author_name": "catadanna",
      "author_url": "",
      "post_date": "05/09/2023 13:20:42",
      "content": "<p>You have three labels here, namely ['StartHesitation', 'Turn' , 'Walking'], and for each label you have several classes (I saw 2 classes for each, 0 and 1, until now, in the training set). For me we have multi class and multi label.</p>\n<p>In this case you have several possibilities : </p>\n<ol>\n<li>Three models : One model for each label (use any classifier)</li>\n<li>Build a neural network and set the output layer with dimension 3.</li>\n</ol>",
      "votes": null,
      "replies": [
        {
          "id": 2251608,
          "author_name": "avivlevi815",
          "author_url": "",
          "post_date": "05/09/2023 14:08:50",
          "content": "<p>I use 4 classes, \"Nothing\" included- to predict there is no event right now.</p>",
          "votes": null,
          "replies": [
            {
              "id": 2251728,
              "author_name": "atamazian",
              "author_url": "",
              "post_date": "05/09/2023 15:31:41",
              "content": "<p>I also use 4 classes in my model.</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2251418": "Am I right that this competition is a multiclass problem, not multilabel one? Meaning that only one class from ['StartHesitation', 'Turn' , 'Walking'] may be assigned to a specific record.",
    "2251552": "You have three labels here, namely ['StartHesitation', 'Turn' , 'Walking'], and for each label you have several classes (I saw 2 classes for each, 0 and 1, until now, in the training set). For me we have multi class and multi label.\n\nIn this case you have several possibilities : \n\n1. Three models : One model for each label (use any classifier)\n2. Build a neural network and set the output layer with dimension 3.",
    "2251608": "I use 4 classes, \"Nothing\" included- to predict there is no event right now.",
    "2251728": "I also use 4 classes in my model."
  },
  "source": "meta"
}