{
  "id": 78429,
  "title": "Feature Selection - Variance Threshold",
  "url": "/competitions/elo-merchant-category-recommendation/discussion/78429",
  "author_name": "Ioannes Paulus",
  "post_date": "2019-01-23T17:09:22.123000",
  "votes": 0,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hi, has anyone there dropped any column based in the variance threshold algorithm or any other? This method dropped some 2 columns with low variance (.8 threshold) in \"new_transactions\", category_1 and 2. Also, I notice that category_3 is just a classification of number of installments 0, 1, and &gt; 1. So I decided to drop it too. Would like to have an idea if I am doing something wrong.</p>",
  "messages": [
    {
      "id": 461324,
      "postDate": "2019-01-25T18:53:40.937Z",
      "content": "<p>Here is an example of recursive feature selection on ELO\n<a href=\"https://www.kaggle.com/roydatascience/recursive-feature-selection-using-sklearn-on-elo?scriptVersionId=9917777\">https://www.kaggle.com/roydatascience/recursive-feature-selection-using-sklearn-on-elo?scriptVersionId=9917777</a></p>",
      "rawMarkdown": "Here is an example of recursive feature selection on ELO\nhttps://www.kaggle.com/roydatascience/recursive-feature-selection-using-sklearn-on-elo?scriptVersionId=9917777",
      "votes": 3
    },
    {
      "id": 460866,
      "postDate": "2019-01-24T15:14:29.627Z",
      "content": "<p>Don't rely on one method for feature selection. You can try some good feature selections method like:\n1) <a href=\"https://www.kaggle.com/tilii7/boruta-feature-elimination\">https://www.kaggle.com/tilii7/boruta-feature-elimination</a>).\n2) <a href=\"https://www.kaggle.com/ogrellier/feature-selection-with-null-importances\">https://www.kaggle.com/ogrellier/feature-selection-with-null-importances</a></p>",
      "rawMarkdown": "Don't rely on one method for feature selection. You can try some good feature selections method like:\n1) https://www.kaggle.com/tilii7/boruta-feature-elimination).\n2) https://www.kaggle.com/ogrellier/feature-selection-with-null-importances",
      "votes": 1,
      "replies": [
        {
          "id": 460920,
          "postDate": "2019-01-24T18:13:22.677Z",
          "content": "<p>Thank you for the links, Harshit. Will take a look.</p>",
          "rawMarkdown": "Thank you for the links, Harshit. Will take a look.",
          "votes": 1
        }
      ]
    },
    {
      "id": 460427,
      "postDate": "2019-01-23T17:09:22.123Z",
      "content": "<p>Hi, has anyone there dropped any column based in the variance threshold algorithm or any other? This method dropped some 2 columns with low variance (.8 threshold) in \"new_transactions\", category_1 and 2. Also, I notice that category_3 is just a classification of number of installments 0, 1, and &gt; 1. So I decided to drop it too. Would like to have an idea if I am doing something wrong.</p>",
      "rawMarkdown": "Hi, has anyone there dropped any column based in the variance threshold algorithm or any other? This method dropped some 2 columns with low variance (.8 threshold) in \"new_transactions\", category_1 and 2. Also, I notice that category_3 is just a classification of number of installments 0, 1, and &gt; 1. So I decided to drop it too. Would like to have an idea if I am doing something wrong."
    },
    {
      "id": 460872,
      "postDate": "2019-01-24T15:22:58.830Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 461324,
      "author_name": "Ashish Gupta",
      "author_url": "",
      "post_date": "2019-01-25T18:53:40.937000",
      "content": "<p>Here is an example of recursive feature selection on ELO\n<a href=\"https://www.kaggle.com/roydatascience/recursive-feature-selection-using-sklearn-on-elo?scriptVersionId=9917777\">https://www.kaggle.com/roydatascience/recursive-feature-selection-using-sklearn-on-elo?scriptVersionId=9917777</a></p>",
      "votes": 3,
      "replies": []
    },
    {
      "id": 460866,
      "author_name": "HarshitMehta",
      "author_url": "",
      "post_date": "2019-01-24T15:14:29.627000",
      "content": "<p>Don't rely on one method for feature selection. You can try some good feature selections method like:\n1) <a href=\"https://www.kaggle.com/tilii7/boruta-feature-elimination\">https://www.kaggle.com/tilii7/boruta-feature-elimination</a>).\n2) <a href=\"https://www.kaggle.com/ogrellier/feature-selection-with-null-importances\">https://www.kaggle.com/ogrellier/feature-selection-with-null-importances</a></p>",
      "votes": 1,
      "replies": [
        {
          "id": 460920,
          "author_name": "Ioannes Paulus",
          "author_url": "",
          "post_date": "2019-01-24T18:13:22.677000",
          "content": "<p>Thank you for the links, Harshit. Will take a look.</p>",
          "votes": 1,
          "replies": []
        }
      ]
    },
    {
      "id": 460872,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-01-24T15:22:58.830000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "461324": "Here is an example of recursive feature selection on ELO\nhttps://www.kaggle.com/roydatascience/recursive-feature-selection-using-sklearn-on-elo?scriptVersionId=9917777",
    "460866": "Don't rely on one method for feature selection. You can try some good feature selections method like:\n1) https://www.kaggle.com/tilii7/boruta-feature-elimination).\n2) https://www.kaggle.com/ogrellier/feature-selection-with-null-importances",
    "460427": "Hi, has anyone there dropped any column based in the variance threshold algorithm or any other? This method dropped some 2 columns with low variance (.8 threshold) in \"new_transactions\", category_1 and 2. Also, I notice that category_3 is just a classification of number of installments 0, 1, and &gt; 1. So I decided to drop it too. Would like to have an idea if I am doing something wrong.",
    "460872": ""
  }
}