{
  "id": 541688,
  "title": "How to solve this issue \"[LightGBM] [Fatal] The number of features in data (14) is not the same as it was in training data (81).\"?",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/541688",
  "author_name": "",
  "post_date": "2024-10-20T22:59:41.829666900Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>How to solve this issue \"[LightGBM] [Fatal] The number of features in data (14) is not the same as it was in training data (81).\"?</p>",
  "messages": [
    {
      "id": "3023731",
      "postDate": "10/20/2024 22:59:41",
      "content": "<p>How to solve this issue \"[LightGBM] [Fatal] The number of features in data (14) is not the same as it was in training data (81).\"?</p>",
      "rawMarkdown": "How to solve this issue \"[LightGBM] [Fatal] The number of features in data (14) is not the same as it was in training data (81).\"?",
      "votes": null
    },
    {
      "id": "3023798",
      "postDate": "10/21/2024 03:15:40",
      "content": "<p>make them consistant.</p>",
      "rawMarkdown": "make them consistant.",
      "votes": null
    },
    {
      "id": "3023805",
      "postDate": "10/21/2024 03:43:38",
      "content": "<p>The error message is already quite straight forward IMHO.</p>\n<p>Whatever preprocessing and feature engineering steps you made for your training data should also be the same for your next set of datasets (validation, test, etc). Though take note of specific transformations like fit_transform vs transform, know which to apply against training and test and validation datasets.</p>\n<p>For the community to best help you, please share your kernel or notebook (though this is a competition) .</p>\n<p>Im sure you'd fix it soon and some of the folks here would help you sort things out.</p>",
      "rawMarkdown": "The error message is already quite straight forward IMHO.\n\nWhatever preprocessing and feature engineering steps you made for your training data should also be the same for your next set of datasets (validation, test, etc). Though take note of specific transformations like fit_transform vs transform, know which to apply against training and test and validation datasets.\n\nFor the community to best help you, please share your kernel or notebook (though this is a competition) .\n\nIm sure you'd fix it soon and some of the folks here would help you sort things out.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3023798,
      "author_name": "dragonzhang",
      "author_url": "",
      "post_date": "10/21/2024 03:15:40",
      "content": "<p>make them consistant.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3023805,
      "author_name": "bwandowando",
      "author_url": "",
      "post_date": "10/21/2024 03:43:38",
      "content": "<p>The error message is already quite straight forward IMHO.</p>\n<p>Whatever preprocessing and feature engineering steps you made for your training data should also be the same for your next set of datasets (validation, test, etc). Though take note of specific transformations like fit_transform vs transform, know which to apply against training and test and validation datasets.</p>\n<p>For the community to best help you, please share your kernel or notebook (though this is a competition) .</p>\n<p>Im sure you'd fix it soon and some of the folks here would help you sort things out.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3023731": "How to solve this issue \"[LightGBM] [Fatal] The number of features in data (14) is not the same as it was in training data (81).\"?",
    "3023798": "make them consistant.",
    "3023805": "The error message is already quite straight forward IMHO.\n\nWhatever preprocessing and feature engineering steps you made for your training data should also be the same for your next set of datasets (validation, test, etc). Though take note of specific transformations like fit_transform vs transform, know which to apply against training and test and validation datasets.\n\nFor the community to best help you, please share your kernel or notebook (though this is a competition) .\n\nIm sure you'd fix it soon and some of the folks here would help you sort things out."
  },
  "source": "meta"
}