{
  "id": 411092,
  "title": "Why do top Notebooks use a Regressor model?",
  "url": "/competitions/tlvmc-parkinsons-freezing-gait-prediction/discussion/411092",
  "author_name": "",
  "post_date": "2023-05-17T17:07:27.242089700Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I was examining some notebooks, and it seems using a regression model is not uncommon.<br>\nAny idea why? I use a classifier model + predict_proba<br>\nIs there any benefit to using a regression model for this task?</p>",
  "messages": [
    {
      "id": "2263551",
      "postDate": "05/17/2023 17:07:27",
      "content": "<p>I was examining some notebooks, and it seems using a regression model is not uncommon.<br>\nAny idea why? I use a classifier model + predict_proba<br>\nIs there any benefit to using a regression model for this task?</p>",
      "rawMarkdown": "I was examining some notebooks, and it seems using a regression model is not uncommon.\nAny idea why? I use a classifier model + predict_proba\nIs there any benefit to using a regression model for this task?",
      "votes": null
    },
    {
      "id": "2265607",
      "postDate": "05/19/2023 11:28:12",
      "content": "<p>I'm also puzzled by the choice of regression models, since it's clearly a multiclass classification problem.</p>",
      "rawMarkdown": "I'm also puzzled by the choice of regression models, since it's clearly a multiclass classification problem.",
      "votes": null
    },
    {
      "id": "2265893",
      "postDate": "05/19/2023 15:27:45",
      "content": "<p>One could argue that qe are predicting values between 0 and 1 so it's a regression problem </p>",
      "rawMarkdown": "One could argue that qe are predicting values between 0 and 1 so it's a regression problem",
      "votes": null
    },
    {
      "id": "2265902",
      "postDate": "05/19/2023 15:32:57",
      "content": "<p>You can get probabilities for each class from any classifier. For example, you can use <code>predict_proba</code> for that in LightGBM.</p>",
      "rawMarkdown": "You can get probabilities for each class from any classifier. For example, you can use `predict_proba` for that in LightGBM.",
      "votes": null
    },
    {
      "id": "2266039",
      "postDate": "05/19/2023 17:00:10",
      "content": "<p>Yap, that's my way too.</p>",
      "rawMarkdown": "Yap, that's my way too.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2265607,
      "author_name": "atamazian",
      "author_url": "",
      "post_date": "05/19/2023 11:28:12",
      "content": "<p>I'm also puzzled by the choice of regression models, since it's clearly a multiclass classification problem.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2265893,
          "author_name": "avivlevi815",
          "author_url": "",
          "post_date": "05/19/2023 15:27:45",
          "content": "<p>One could argue that qe are predicting values between 0 and 1 so it's a regression problem </p>",
          "votes": null,
          "replies": [
            {
              "id": 2265902,
              "author_name": "atamazian",
              "author_url": "",
              "post_date": "05/19/2023 15:32:57",
              "content": "<p>You can get probabilities for each class from any classifier. For example, you can use <code>predict_proba</code> for that in LightGBM.</p>",
              "votes": null,
              "replies": [
                {
                  "id": 2266039,
                  "author_name": "avivlevi815",
                  "author_url": "",
                  "post_date": "05/19/2023 17:00:10",
                  "content": "<p>Yap, that's my way too.</p>",
                  "votes": null,
                  "replies": []
                }
              ]
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2263551": "I was examining some notebooks, and it seems using a regression model is not uncommon.\nAny idea why? I use a classifier model + predict_proba\nIs there any benefit to using a regression model for this task?",
    "2265607": "I'm also puzzled by the choice of regression models, since it's clearly a multiclass classification problem.",
    "2265893": "One could argue that qe are predicting values between 0 and 1 so it's a regression problem",
    "2265902": "You can get probabilities for each class from any classifier. For example, you can use `predict_proba` for that in LightGBM.",
    "2266039": "Yap, that's my way too."
  },
  "source": "meta"
}