{
  "id": 105299,
  "title": "Interesting negative LB score",
  "url": "/competitions/aptos2019-blindness-detection/discussion/105299",
  "author_name": "Levis",
  "post_date": "2019-08-22T07:54:53.555000",
  "votes": 0,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hey guys, my model LB score is **-0.128. **Why does this happen?What are its possible causes?</p>",
  "messages": [
    {
      "id": 605491,
      "postDate": "2019-08-22T12:22:07.567Z",
      "content": "<p>qappa metric has a range of -1 to +1. If your model predicts class 1 data as class 5 then qappa lowers compared to if your model predicts class 1 data as class 2. Hope this explanation helps.</p>",
      "rawMarkdown": "qappa metric has a range of -1 to +1. If your model predicts class 1 data as class 5 then qappa lowers compared to if your model predicts class 1 data as class 2. Hope this explanation helps.",
      "votes": 1,
      "replies": [
        {
          "id": 611364,
          "postDate": "2019-08-29T08:20:42.210Z",
          "content": "<p>thanks a lot !👌 </p>",
          "rawMarkdown": "thanks a lot !👌 "
        }
      ]
    },
    {
      "id": 605267,
      "postDate": "2019-08-22T07:54:53.557Z",
      "content": "<p>Hey guys, my model LB score is **-0.128. **Why does this happen?What are its possible causes?</p>",
      "rawMarkdown": "Hey guys, my model LB score is **-0.128. **Why does this happen?What are its possible causes?"
    },
    {
      "id": 605587,
      "postDate": "2019-08-22T14:14:22.050Z",
      "rawMarkdown": "",
      "isDeleted": true,
      "replies": [
        {
          "id": 611362,
          "postDate": "2019-08-29T08:20:19.830Z",
          "content": "<p>thanks a lot !👍 </p>",
          "rawMarkdown": "thanks a lot !👍 "
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 605491,
      "author_name": "Sabbir Ahmed",
      "author_url": "",
      "post_date": "2019-08-22T12:22:07.567000",
      "content": "<p>qappa metric has a range of -1 to +1. If your model predicts class 1 data as class 5 then qappa lowers compared to if your model predicts class 1 data as class 2. Hope this explanation helps.</p>",
      "votes": 1,
      "replies": [
        {
          "id": 611364,
          "author_name": "Levis",
          "author_url": "",
          "post_date": "2019-08-29T08:20:42.210000",
          "content": "<p>thanks a lot !👌 </p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 605587,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-08-22T14:14:22.050000",
      "content": "",
      "votes": 0,
      "replies": [
        {
          "id": 611362,
          "author_name": "Levis",
          "author_url": "",
          "post_date": "2019-08-29T08:20:19.830000",
          "content": "<p>thanks a lot !👍 </p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "605491": "qappa metric has a range of -1 to +1. If your model predicts class 1 data as class 5 then qappa lowers compared to if your model predicts class 1 data as class 2. Hope this explanation helps.",
    "605267": "Hey guys, my model LB score is **-0.128. **Why does this happen?What are its possible causes?",
    "605587": ""
  }
}