{
  "id": 314151,
  "title": "How to determine the threshold for classification?",
  "url": "/competitions/birdclef-2022/discussion/314151",
  "author_name": "",
  "post_date": "2022-03-21T08:35:11.505879900Z",
  "votes": 6,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Since the test data distribution is not visible and the classes are extremely imbalanced, it can be difficult to determine the threshold for a classification model. </p>\n<p>Could anyone share some ideas on how to find a legit threshold? Thank you! </p>",
  "messages": [
    {
      "id": "1730424",
      "postDate": "03/21/2022 08:35:11",
      "content": "<p>Since the test data distribution is not visible and the classes are extremely imbalanced, it can be difficult to determine the threshold for a classification model. </p>\n<p>Could anyone share some ideas on how to find a legit threshold? Thank you! </p>",
      "rawMarkdown": "Since the test data distribution is not visible and the classes are extremely imbalanced, it can be difficult to determine the threshold for a classification model. \n\nCould anyone share some ideas on how to find a legit threshold? Thank you!",
      "votes": null
    },
    {
      "id": "1732471",
      "postDate": "03/23/2022 12:21:31",
      "content": "<p>what is threshold means? if you ask me the imbalanced in classification issue, I use the Data_Augmentation to solve that.</p>",
      "rawMarkdown": "what is threshold means? if you ask me the imbalanced in classification issue, I use the Data_Augmentation to solve that.",
      "votes": null
    },
    {
      "id": "1752604",
      "postDate": "04/12/2022 01:09:59",
      "content": "<p>You just need to output probabilities. You dont need to calculate threshold.</p>",
      "rawMarkdown": "You just need to output probabilities. You dont need to calculate threshold.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1732471,
      "author_name": "yoyobar",
      "author_url": "",
      "post_date": "03/23/2022 12:21:31",
      "content": "<p>what is threshold means? if you ask me the imbalanced in classification issue, I use the Data_Augmentation to solve that.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1752604,
      "author_name": "harsha1999",
      "author_url": "",
      "post_date": "04/12/2022 01:09:59",
      "content": "<p>You just need to output probabilities. You dont need to calculate threshold.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1730424": "Since the test data distribution is not visible and the classes are extremely imbalanced, it can be difficult to determine the threshold for a classification model. \n\nCould anyone share some ideas on how to find a legit threshold? Thank you!",
    "1732471": "what is threshold means? if you ask me the imbalanced in classification issue, I use the Data_Augmentation to solve that.",
    "1752604": "You just need to output probabilities. You dont need to calculate threshold."
  },
  "source": "meta"
}