{
  "id": 77478,
  "title": "How to deal with the missing values?",
  "url": "/competitions/elo-merchant-category-recommendation/discussion/77478",
  "author_name": "",
  "post_date": "2019-01-13T08:02:14.800025500Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi, everyone! As you may found that there are a lot of missing values in the competition data, such as <code>np.nan</code> in <code>category_2/3</code>, <code>-1</code> in <code>*_id</code>, etc. So, I am wondering that what should I do to deal with those missing values. As far as I konw, there are three main approachs: 1. filling with mod or mean values; 2. label it into an another category; 3. drop.\nSo, which way is the best and what should I chose  to do for this specific competition?</p>",
  "messages": [
    {
      "id": "455197",
      "postDate": "01/13/2019 08:02:14",
      "content": "<p>Hi, everyone! As you may found that there are a lot of missing values in the competition data, such as <code>np.nan</code> in <code>category_2/3</code>, <code>-1</code> in <code>*_id</code>, etc. So, I am wondering that what should I do to deal with those missing values. As far as I konw, there are three main approachs: 1. filling with mod or mean values; 2. label it into an another category; 3. drop.\nSo, which way is the best and what should I chose  to do for this specific competition?</p>",
      "rawMarkdown": "Hi, everyone! As you may found that there are a lot of missing values in the competition data, such as `np.nan` in `category_2/3`, `-1` in `*_id`, etc. So, I am wondering that what should I do to deal with those missing values. As far as I konw, there are three main approachs: 1. filling with mod or mean values; 2. label it into an another category; 3. drop.\nSo, which way is the best and what should I chose  to do for this specific competition?",
      "votes": null
    },
    {
      "id": "455349",
      "postDate": "01/13/2019 16:38:46",
      "content": "<p>i have filed numerical features missing values with the mode as well as mean. Like in case of no of floors, garages etc in that i have used mode and other like area column in that i have used mean.\nIn categorical features i have used none. Dropping is not a good idea. I have dropped only those with outliers.</p>",
      "rawMarkdown": "i have filed numerical features missing values with the mode as well as mean. Like in case of no of floors, garages etc in that i have used mode and other like area column in that i have used mean.\nIn categorical features i have used none. Dropping is not a good idea. I have dropped only those with outliers.",
      "votes": null
    },
    {
      "id": "455738",
      "postDate": "01/14/2019 13:21:09",
      "content": "<p>Thank you very much. I don't konw if I make it clear(o(╥﹏╥)o my poor English). I guess you have filled the numerical features with mean values, and label categorical features into a new category</p>",
      "rawMarkdown": "Thank you very much. I don't konw if I make it clear(o(╥﹏╥)o my poor English). I guess you have filled the numerical features with mean values, and label categorical features into a new category",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 455349,
      "author_name": "alokpratap",
      "author_url": "",
      "post_date": "01/13/2019 16:38:46",
      "content": "<p>i have filed numerical features missing values with the mode as well as mean. Like in case of no of floors, garages etc in that i have used mode and other like area column in that i have used mean.\nIn categorical features i have used none. Dropping is not a good idea. I have dropped only those with outliers.</p>",
      "votes": null,
      "replies": [
        {
          "id": 455738,
          "author_name": "liewmanchoi",
          "author_url": "",
          "post_date": "01/14/2019 13:21:09",
          "content": "<p>Thank you very much. I don't konw if I make it clear(o(╥﹏╥)o my poor English). I guess you have filled the numerical features with mean values, and label categorical features into a new category</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "455197": "Hi, everyone! As you may found that there are a lot of missing values in the competition data, such as `np.nan` in `category_2/3`, `-1` in `*_id`, etc. So, I am wondering that what should I do to deal with those missing values. As far as I konw, there are three main approachs: 1. filling with mod or mean values; 2. label it into an another category; 3. drop.\nSo, which way is the best and what should I chose  to do for this specific competition?",
    "455349": "i have filed numerical features missing values with the mode as well as mean. Like in case of no of floors, garages etc in that i have used mode and other like area column in that i have used mean.\nIn categorical features i have used none. Dropping is not a good idea. I have dropped only those with outliers.",
    "455738": "Thank you very much. I don't konw if I make it clear(o(╥﹏╥)o my poor English). I guess you have filled the numerical features with mean values, and label categorical features into a new category"
  },
  "source": "meta"
}