{
  "id": 463921,
  "title": "Doesn’t the hgsc category account for more than 60% in this competition? If the model classifies the image as hgsc when it sees the image, then the accuracy will reach 0.6? Or is it judged by the average accuracy of each category?",
  "url": "/competitions/UBC-OCEAN/discussion/463921",
  "author_name": "Metavers",
  "post_date": "2023-12-28T01:32:19.493000",
  "votes": 2,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Doesn’t the hgsc category account for more than 60% in this competition? If the model classifies the image as hgsc when it sees the image, then the accuracy will reach 0.6? Or is it judged by the average accuracy of each category?</p>",
  "messages": [
    {
      "id": 2578091,
      "postDate": "2023-12-29T03:37:46.823Z",
      "content": "<p>the test metrics is more like average recall than accuracy so (1+0+0+0+0+0)/6=0.16</p>",
      "rawMarkdown": "the test metrics is more like average recall than accuracy so (1+0+0+0+0+0)/6=0.16",
      "votes": 3
    },
    {
      "id": 2584014,
      "postDate": "2024-01-02T16:26:56.717Z",
      "content": "<p>The metric is balanced accuracy, meaning it will calculate the accuracy of each label and then take the unweighted mean.  For example, if we predict HGSC every time, we will have a 1.0 accuracy for that label, but a 0.0 for all the others, giving a balanced accuracy of (1.0 + 0.0 + 0.0 + 0.0 + 0.0 + 0.0) / 6</p>",
      "rawMarkdown": "The metric is balanced accuracy, meaning it will calculate the accuracy of each label and then take the unweighted mean.  For example, if we predict HGSC every time, we will have a 1.0 accuracy for that label, but a 0.0 for all the others, giving a balanced accuracy of (1.0 + 0.0 + 0.0 + 0.0 + 0.0 + 0.0) / 6",
      "votes": 1
    },
    {
      "id": 2576646,
      "postDate": "2023-12-28T01:32:19.493Z",
      "content": "<p>Doesn’t the hgsc category account for more than 60% in this competition? If the model classifies the image as hgsc when it sees the image, then the accuracy will reach 0.6? Or is it judged by the average accuracy of each category?</p>",
      "rawMarkdown": "Doesn’t the hgsc category account for more than 60% in this competition? If the model classifies the image as hgsc when it sees the image, then the accuracy will reach 0.6? Or is it judged by the average accuracy of each category?",
      "votes": 2
    },
    {
      "id": 2576837,
      "postDate": "2023-12-28T06:49:24.657Z",
      "content": "<p>Metric is Balanced Accuracy, so it will give you 0.17 score ig</p>",
      "rawMarkdown": "Metric is Balanced Accuracy, so it will give you 0.17 score ig"
    }
  ],
  "comments": [
    {
      "id": 2578091,
      "author_name": "Yoobin",
      "author_url": "",
      "post_date": "2023-12-29T03:37:46.823000",
      "content": "<p>the test metrics is more like average recall than accuracy so (1+0+0+0+0+0)/6=0.16</p>",
      "votes": 3,
      "replies": []
    },
    {
      "id": 2584014,
      "author_name": "Connor",
      "author_url": "",
      "post_date": "2024-01-02T16:26:56.717000",
      "content": "<p>The metric is balanced accuracy, meaning it will calculate the accuracy of each label and then take the unweighted mean.  For example, if we predict HGSC every time, we will have a 1.0 accuracy for that label, but a 0.0 for all the others, giving a balanced accuracy of (1.0 + 0.0 + 0.0 + 0.0 + 0.0 + 0.0) / 6</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 2576837,
      "author_name": "继续战斗继续前进",
      "author_url": "",
      "post_date": "2023-12-28T06:49:24.657000",
      "content": "<p>Metric is Balanced Accuracy, so it will give you 0.17 score ig</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2578091": "the test metrics is more like average recall than accuracy so (1+0+0+0+0+0)/6=0.16",
    "2584014": "The metric is balanced accuracy, meaning it will calculate the accuracy of each label and then take the unweighted mean.  For example, if we predict HGSC every time, we will have a 1.0 accuracy for that label, but a 0.0 for all the others, giving a balanced accuracy of (1.0 + 0.0 + 0.0 + 0.0 + 0.0 + 0.0) / 6",
    "2576646": "Doesn’t the hgsc category account for more than 60% in this competition? If the model classifies the image as hgsc when it sees the image, then the accuracy will reach 0.6? Or is it judged by the average accuracy of each category?",
    "2576837": "Metric is Balanced Accuracy, so it will give you 0.17 score ig"
  }
}