{
  "id": 651804,
  "title": "Low Score even with good training score",
  "url": "/competitions/MABe-mouse-behavior-detection/discussion/651804",
  "author_name": "",
  "post_date": "2025-12-05T10:33:31.410890700Z",
  "votes": -1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi all, I have trained a Light GBM model on labeled data (843/848). F1 score on the unseen data (5/848) is 0.65, when I submit my predictions on the test file I got a score of 0.006 only. I have formatted the submission file as required (sample is given below). Can someone help me why is my score so low? when it got 0.65 on unseen data, arguably test data is also will be similar if not 0.65 I was expecting something like 0.2-0.3. </p>\n<p>I have excluded unlabeled data for the initial run. </p>\n<p>Sample submission is attached</p>\n<p>Help me out if someone knows what I should do!!!</p>",
  "messages": [
    {
      "id": "3362799",
      "postDate": "12/05/2025 10:33:31",
      "content": "<p>Hi all, I have trained a Light GBM model on labeled data (843/848). F1 score on the unseen data (5/848) is 0.65, when I submit my predictions on the test file I got a score of 0.006 only. I have formatted the submission file as required (sample is given below). Can someone help me why is my score so low? when it got 0.65 on unseen data, arguably test data is also will be similar if not 0.65 I was expecting something like 0.2-0.3. </p>\n<p>I have excluded unlabeled data for the initial run. </p>\n<p>Sample submission is attached</p>\n<p>Help me out if someone knows what I should do!!!</p>",
      "rawMarkdown": "Hi all, I have trained a Light GBM model on labeled data (843/848). F1 score on the unseen data (5/848) is 0.65, when I submit my predictions on the test file I got a score of 0.006 only. I have formatted the submission file as required (sample is given below). Can someone help me why is my score so low? when it got 0.65 on unseen data, arguably test data is also will be similar if not 0.65 I was expecting something like 0.2-0.3. \n\nI have excluded unlabeled data for the initial run. \n\nSample submission is attached\n\nHelp me out if someone knows what I should do!!!",
      "votes": null
    },
    {
      "id": "3362838",
      "postDate": "12/05/2025 11:59:52",
      "content": "<p>First, make sure your f1 score matches the competition metric: <a href=\"https://www.kaggle.com/code/metric/mabe-f-beta\" target=\"_blank\">the full implementation here</a></p>\n<p>Second, the test set will be replaced by a hiddent test set, so you must include code to load models, features extraction and make submission file on the submitted notebook</p>",
      "rawMarkdown": "First, make sure your f1 score matches the competition metric: [the full implementation here](https://www.kaggle.com/code/metric/mabe-f-beta)\n\nSecond, the test set will be replaced by a hiddent test set, so you must include code to load models, features extraction and make submission file on the submitted notebook",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3362838,
      "author_name": "llkh0a",
      "author_url": "",
      "post_date": "12/05/2025 11:59:52",
      "content": "<p>First, make sure your f1 score matches the competition metric: <a href=\"https://www.kaggle.com/code/metric/mabe-f-beta\" target=\"_blank\">the full implementation here</a></p>\n<p>Second, the test set will be replaced by a hiddent test set, so you must include code to load models, features extraction and make submission file on the submitted notebook</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3362799": "Hi all, I have trained a Light GBM model on labeled data (843/848). F1 score on the unseen data (5/848) is 0.65, when I submit my predictions on the test file I got a score of 0.006 only. I have formatted the submission file as required (sample is given below). Can someone help me why is my score so low? when it got 0.65 on unseen data, arguably test data is also will be similar if not 0.65 I was expecting something like 0.2-0.3. \n\nI have excluded unlabeled data for the initial run. \n\nSample submission is attached\n\nHelp me out if someone knows what I should do!!!",
    "3362838": "First, make sure your f1 score matches the competition metric: [the full implementation here](https://www.kaggle.com/code/metric/mabe-f-beta)\n\nSecond, the test set will be replaced by a hiddent test set, so you must include code to load models, features extraction and make submission file on the submitted notebook"
  },
  "source": "meta"
}