{
  "id": 176872,
  "title": "segment-level prediction required, about 150 segment-level annotated, only 2 available?",
  "url": "/competitions/birdsong-recognition/discussion/176872",
  "author_name": "yama",
  "post_date": "2020-08-23T23:06:58.322000",
  "votes": 3,
  "comment_count": 0,
  "views": 0,
  "content": "<p>It means that train : test ~= 2 : 150.<br>\nI think that the split ratio is unnatural and impractical.</p>\n<p>There are too few samples to evaluate the metric locally (especially negative samples).</p>\n<p>How do you think about it?<br>\nAm I missing something?</p>",
  "messages": [
    {
      "id": 982984,
      "postDate": "2020-08-23T23:06:58.323Z",
      "content": "<p>It means that train : test ~= 2 : 150.<br>\nI think that the split ratio is unnatural and impractical.</p>\n<p>There are too few samples to evaluate the metric locally (especially negative samples).</p>\n<p>How do you think about it?<br>\nAm I missing something?</p>",
      "rawMarkdown": "It means that train : test ~= 2 : 150.\nI think that the split ratio is unnatural and impractical.\n\nThere are too few samples to evaluate the metric locally (especially negative samples).\n\nHow do you think about it?\nAm I missing something?",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "982984": "It means that train : test ~= 2 : 150.\nI think that the split ratio is unnatural and impractical.\n\nThere are too few samples to evaluate the metric locally (especially negative samples).\n\nHow do you think about it?\nAm I missing something?"
  }
}