{
  "id": 541200,
  "title": "How to Fill NaN?",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/541200",
  "author_name": "",
  "post_date": "2024-10-18T07:07:25.959006400Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n<p>Hope you all enjoy the competition.</p>\n<p>I'm now exploring the top ranking solutions of the previous JS competition. I saw in <a href=\"https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224713\" target=\"_blank\">3rd place solution</a>, he/she/they/non-binary mentioned that <code>NaN filling: median conditioned on f0</code>. I'm a little confused what it means. Any one knows what <code>f0</code> is? Or do you have any NaN filling strats to share/recommend?</p>\n<p>Thanks in advance!</p>",
  "messages": [
    {
      "id": "3021072",
      "postDate": "10/18/2024 07:07:25",
      "content": "<p>Hi everyone,</p>\n<p>Hope you all enjoy the competition.</p>\n<p>I'm now exploring the top ranking solutions of the previous JS competition. I saw in <a href=\"https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224713\" target=\"_blank\">3rd place solution</a>, he/she/they/non-binary mentioned that <code>NaN filling: median conditioned on f0</code>. I'm a little confused what it means. Any one knows what <code>f0</code> is? Or do you have any NaN filling strats to share/recommend?</p>\n<p>Thanks in advance!</p>",
      "rawMarkdown": "Hi everyone,\n\nHope you all enjoy the competition.\n\nI'm now exploring the top ranking solutions of the previous JS competition. I saw in [3rd place solution](https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224713), he/she/they/non-binary mentioned that `NaN filling: median conditioned on f0`. I'm a little confused what it means. Any one knows what `f0` is? Or do you have any NaN filling strats to share/recommend?\n\nThanks in advance!",
      "votes": null
    },
    {
      "id": "3022958",
      "postDate": "10/20/2024 05:55:08",
      "content": "<p>Based on what i can see, i think rather than filling with median value of whole column, what they did is they divided the data in subgroup and filled the NaN values with corresponding group's median.<br>\n(Since the data is not available, I can be wrong, but this is what it generally means.)</p>",
      "rawMarkdown": "Based on what i can see, i think rather than filling with median value of whole column, what they did is they divided the data in subgroup and filled the NaN values with corresponding group's median.\n(Since the data is not available, I can be wrong, but this is what it generally means.)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3022958,
      "author_name": "jayshrivastava",
      "author_url": "",
      "post_date": "10/20/2024 05:55:08",
      "content": "<p>Based on what i can see, i think rather than filling with median value of whole column, what they did is they divided the data in subgroup and filled the NaN values with corresponding group's median.<br>\n(Since the data is not available, I can be wrong, but this is what it generally means.)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3021072": "Hi everyone,\n\nHope you all enjoy the competition.\n\nI'm now exploring the top ranking solutions of the previous JS competition. I saw in [3rd place solution](https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224713), he/she/they/non-binary mentioned that `NaN filling: median conditioned on f0`. I'm a little confused what it means. Any one knows what `f0` is? Or do you have any NaN filling strats to share/recommend?\n\nThanks in advance!",
    "3022958": "Based on what i can see, i think rather than filling with median value of whole column, what they did is they divided the data in subgroup and filled the NaN values with corresponding group's median.\n(Since the data is not available, I can be wrong, but this is what it generally means.)"
  },
  "source": "meta"
}