{
  "id": 621844,
  "title": "What does “minus” even mean??",
  "url": "/competitions/csiro-biomass/discussion/621844",
  "author_name": "",
  "post_date": "2025-11-15T10:59:33.714786900Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I just submitted to a real Kaggle competition for the very first time!\nI’m so excited!\nI still don’t understand machine learning at all — I asked ChatGPT about everything and submitted the code this time.\nI’m going to keep studying from here.</p>\n<p>By the way, what does it mean when the score is negative!?\nI thought the lowest possible score would be 0.00, haha.</p>",
  "messages": [
    {
      "id": "3326579",
      "postDate": "11/15/2025 10:59:33",
      "content": "<p>I just submitted to a real Kaggle competition for the very first time!\nI’m so excited!\nI still don’t understand machine learning at all — I asked ChatGPT about everything and submitted the code this time.\nI’m going to keep studying from here.</p>\n<p>By the way, what does it mean when the score is negative!?\nI thought the lowest possible score would be 0.00, haha.</p>",
      "rawMarkdown": "I just submitted to a real Kaggle competition for the very first time!\nI’m so excited!\nI still don’t understand machine learning at all — I asked ChatGPT about everything and submitted the code this time.\nI’m going to keep studying from here.\n\nBy the way, what does it mean when the score is negative!?\nI thought the lowest possible score would be 0.00, haha.",
      "votes": null
    },
    {
      "id": "3327866",
      "postDate": "11/15/2025 16:12:26",
      "content": "<p>If you simply predict the means for the 5 targets for this metric , you get a score around 0.0.  When you get a minus value for the score it simply means that your model is worse than that simple prediction.  If your not running enough epochs you will often see a minus value for the LB score.   So - first simple step - run more epochs for training.  </p>\n<p>In this competion since the R2 values are weighted simple means it generates a lower value.  Often kaggle includes a submission with means - look on the leader board for submission.csv.  That's the score that a simple means prediction will make.     PS - not 100% sure it's the means, it might be the medians!</p>",
      "rawMarkdown": "If you simply predict the means for the 5 targets for this metric , you get a score around 0.0.  When you get a minus value for the score it simply means that your model is worse than that simple prediction.  If your not running enough epochs you will often see a minus value for the LB score.   So - first simple step - run more epochs for training.  \n\nIn this competion since the R2 values are weighted simple means it generates a lower value.  Often kaggle includes a submission with means - look on the leader board for submission.csv.  That's the score that a simple means prediction will make.     PS - not 100% sure it's the means, it might be the medians!",
      "votes": null
    },
    {
      "id": "3331696",
      "postDate": "11/16/2025 13:59:14",
      "content": "<p>Thank you so much for the clear explanation!\nI really appreciate your advice.\nI'll try it out!</p>",
      "rawMarkdown": "Thank you so much for the clear explanation!\nI really appreciate your advice.\nI'll try it out!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3327866,
      "author_name": "pcjimmmy",
      "author_url": "",
      "post_date": "11/15/2025 16:12:26",
      "content": "<p>If you simply predict the means for the 5 targets for this metric , you get a score around 0.0.  When you get a minus value for the score it simply means that your model is worse than that simple prediction.  If your not running enough epochs you will often see a minus value for the LB score.   So - first simple step - run more epochs for training.  </p>\n<p>In this competion since the R2 values are weighted simple means it generates a lower value.  Often kaggle includes a submission with means - look on the leader board for submission.csv.  That's the score that a simple means prediction will make.     PS - not 100% sure it's the means, it might be the medians!</p>",
      "votes": null,
      "replies": [
        {
          "id": 3331696,
          "author_name": "joshuafukasawa",
          "author_url": "",
          "post_date": "11/16/2025 13:59:14",
          "content": "<p>Thank you so much for the clear explanation!\nI really appreciate your advice.\nI'll try it out!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3326579": "I just submitted to a real Kaggle competition for the very first time!\nI’m so excited!\nI still don’t understand machine learning at all — I asked ChatGPT about everything and submitted the code this time.\nI’m going to keep studying from here.\n\nBy the way, what does it mean when the score is negative!?\nI thought the lowest possible score would be 0.00, haha.",
    "3327866": "If you simply predict the means for the 5 targets for this metric , you get a score around 0.0.  When you get a minus value for the score it simply means that your model is worse than that simple prediction.  If your not running enough epochs you will often see a minus value for the LB score.   So - first simple step - run more epochs for training.  \n\nIn this competion since the R2 values are weighted simple means it generates a lower value.  Often kaggle includes a submission with means - look on the leader board for submission.csv.  That's the score that a simple means prediction will make.     PS - not 100% sure it's the means, it might be the medians!",
    "3331696": "Thank you so much for the clear explanation!\nI really appreciate your advice.\nI'll try it out!"
  },
  "source": "meta"
}