{
  "id": 493943,
  "title": "How to effectively perform feature selection and gain benefits? 😁",
  "url": "/competitions/home-credit-credit-risk-model-stability/discussion/493943",
  "author_name": "",
  "post_date": "2024-04-15T13:10:36.434292300Z",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello, kagglers：<br>\n    I am a beginner. I have a question is, apart from the \"reduce_group\" method mentioned in public notebooks, what other approaches can be used?  I had tried like IV(Information Value) and Variance Filter，but it seems not so good?</p>",
  "messages": [
    {
      "id": "2753379",
      "postDate": "04/15/2024 13:10:36",
      "content": "<p>Hello, kagglers：<br>\n    I am a beginner. I have a question is, apart from the \"reduce_group\" method mentioned in public notebooks, what other approaches can be used?  I had tried like IV(Information Value) and Variance Filter，but it seems not so good?</p>",
      "rawMarkdown": "Hello, kagglers：\n    I am a beginner. I have a question is, apart from the \"reduce_group\" method mentioned in public notebooks, what other approaches can be used?  I had tried like IV(Information Value) and Variance Filter，but it seems not so good?",
      "votes": null
    },
    {
      "id": "2753417",
      "postDate": "04/15/2024 13:29:37",
      "content": "<p>Hello! Besides the methods you've tried, consider using Recursive Feature Elimination (RFE) or utilizing tree-based models like Random Forest for their inherent feature importance rankings, which can be very effective for feature selection. Happy kaggling! <a href=\"https://www.kaggle.com/huangshibao\" target=\"_blank\">@huangshibao</a> </p>",
      "rawMarkdown": "Hello! Besides the methods you've tried, consider using Recursive Feature Elimination (RFE) or utilizing tree-based models like Random Forest for their inherent feature importance rankings, which can be very effective for feature selection. Happy kaggling! @huangshibao",
      "votes": null
    },
    {
      "id": "2754323",
      "postDate": "04/16/2024 01:02:35",
      "content": "<p>Thank you for your response. I will give it a try, but I feel that this approach may result in timeouts or out-of-memory errors for a large number of features and data with kaggle notebook.</p>",
      "rawMarkdown": "Thank you for your response. I will give it a try, but I feel that this approach may result in timeouts or out-of-memory errors for a large number of features and data with kaggle notebook.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2753417,
      "author_name": "matinmahmoudi",
      "author_url": "",
      "post_date": "04/15/2024 13:29:37",
      "content": "<p>Hello! Besides the methods you've tried, consider using Recursive Feature Elimination (RFE) or utilizing tree-based models like Random Forest for their inherent feature importance rankings, which can be very effective for feature selection. Happy kaggling! <a href=\"https://www.kaggle.com/huangshibao\" target=\"_blank\">@huangshibao</a> </p>",
      "votes": null,
      "replies": [
        {
          "id": 2754323,
          "author_name": "huangshibao",
          "author_url": "",
          "post_date": "04/16/2024 01:02:35",
          "content": "<p>Thank you for your response. I will give it a try, but I feel that this approach may result in timeouts or out-of-memory errors for a large number of features and data with kaggle notebook.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2753379": "Hello, kagglers：\n    I am a beginner. I have a question is, apart from the \"reduce_group\" method mentioned in public notebooks, what other approaches can be used?  I had tried like IV(Information Value) and Variance Filter，but it seems not so good?",
    "2753417": "Hello! Besides the methods you've tried, consider using Recursive Feature Elimination (RFE) or utilizing tree-based models like Random Forest for their inherent feature importance rankings, which can be very effective for feature selection. Happy kaggling! @huangshibao",
    "2754323": "Thank you for your response. I will give it a try, but I feel that this approach may result in timeouts or out-of-memory errors for a large number of features and data with kaggle notebook."
  },
  "source": "meta"
}