{
  "id": 548986,
  "title": "Does imputation actually help?",
  "url": "/competitions/child-mind-institute-problematic-internet-use/discussion/548986",
  "author_name": "MJeremy",
  "post_date": "2024-11-30T05:17:24.012000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>With the missing values, by using the KNN imputer, given such small data set, many imputed values seem to create actual noise instead of good guesses.</p>\n<p>Example being I tried ones without time series features and pure imputation, the LGB model stops learning at very beginning iterations.</p>",
  "messages": [
    {
      "id": 3058875,
      "postDate": "2024-11-30T05:17:24.013Z",
      "content": "<p>With the missing values, by using the KNN imputer, given such small data set, many imputed values seem to create actual noise instead of good guesses.</p>\n<p>Example being I tried ones without time series features and pure imputation, the LGB model stops learning at very beginning iterations.</p>",
      "rawMarkdown": "With the missing values, by using the KNN imputer, given such small data set, many imputed values seem to create actual noise instead of good guesses.\n\nExample being I tried ones without time series features and pure imputation, the LGB model stops learning at very beginning iterations.",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3058875": "With the missing values, by using the KNN imputer, given such small data set, many imputed values seem to create actual noise instead of good guesses.\n\nExample being I tried ones without time series features and pure imputation, the LGB model stops learning at very beginning iterations."
  }
}