{
  "id": 41474,
  "title": "Gender Column",
  "url": "/competitions/kkbox-churn-prediction-challenge/discussion/41474",
  "author_name": "",
  "post_date": "2017-10-18T15:00:15.870442100Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Since the gender column contains too many missing values in both train and test dataset (around 50%) and the one's that have the value, the distribution is almost same between male and female, so would it be fair to simply drop this column from the predictions.</p>",
  "messages": [
    {
      "id": "232832",
      "postDate": "10/18/2017 15:00:15",
      "content": "<p>Since the gender column contains too many missing values in both train and test dataset (around 50%) and the one's that have the value, the distribution is almost same between male and female, so would it be fair to simply drop this column from the predictions.</p>",
      "rawMarkdown": "Since the gender column contains too many missing values in both train and test dataset (around 50%) and the one's that have the value, the distribution is almost same between male and female, so would it be fair to simply drop this column from the predictions.",
      "votes": null
    },
    {
      "id": "234725",
      "postDate": "10/23/2017 22:38:40",
      "content": "<p>Perhaps set those missing values as \"Not Specified\" could be better. That just my first guess. You can try it.</p>",
      "rawMarkdown": "Perhaps set those missing values as \"Not Specified\" could be better. That just my first guess. You can try it.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 234725,
      "author_name": "jiayuzhang",
      "author_url": "",
      "post_date": "10/23/2017 22:38:40",
      "content": "<p>Perhaps set those missing values as \"Not Specified\" could be better. That just my first guess. You can try it.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "232832": "Since the gender column contains too many missing values in both train and test dataset (around 50%) and the one's that have the value, the distribution is almost same between male and female, so would it be fair to simply drop this column from the predictions.",
    "234725": "Perhaps set those missing values as \"Not Specified\" could be better. That just my first guess. You can try it."
  },
  "source": "meta"
}