{
  "id": 201429,
  "title": "What is songtype? ",
  "url": "/competitions/rfcx-species-audio-detection/discussion/201429",
  "author_name": "",
  "post_date": "2020-12-05T00:58:50.676644700Z",
  "votes": 4,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi, </p>\n<p>I don't understand what to make of the songtype feature in the dataset? I am guess that it describes the different tones made by the same species. However, in that case, it seems that songtype should be taken as label instead of data, so should our model also learn songtype (not required by the contest) as well ?</p>",
  "messages": [
    {
      "id": "1102466",
      "postDate": "12/05/2020 00:58:50",
      "content": "<p>Hi, </p>\n<p>I don't understand what to make of the songtype feature in the dataset? I am guess that it describes the different tones made by the same species. However, in that case, it seems that songtype should be taken as label instead of data, so should our model also learn songtype (not required by the contest) as well ?</p>",
      "rawMarkdown": "Hi, \n\nI don't understand what to make of the songtype feature in the dataset? I am guess that it describes the different tones made by the same species. However, in that case, it seems that songtype should be taken as label instead of data, so should our model also learn songtype (not required by the contest) as well ?",
      "votes": null
    },
    {
      "id": "1102570",
      "postDate": "12/05/2020 04:26:07",
      "content": "<p>You're correct, it's a different tone being sung by the same species. One approach, as you suggest, could be to predict the song, not the species, and <strong>then</strong> use that information to help predict what species it is.</p>",
      "rawMarkdown": "You're correct, it's a different tone being sung by the same species. One approach, as you suggest, could be to predict the song, not the species, and **then** use that information to help predict what species it is.",
      "votes": null
    },
    {
      "id": "1103088",
      "postDate": "12/05/2020 16:12:18",
      "content": "<p>Awesome. Thank you.</p>",
      "rawMarkdown": "Awesome. Thank you.",
      "votes": null
    },
    {
      "id": "1122107",
      "postDate": "12/22/2020 07:38:52",
      "content": "<p>also in submission.csv file they are one hot encoded 1 -&gt; s1 2-&gt;s2</p>",
      "rawMarkdown": "also in submission.csv file they are one hot encoded 1 -> s1 2->s2",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1102570,
      "author_name": "maltonji",
      "author_url": "",
      "post_date": "12/05/2020 04:26:07",
      "content": "<p>You're correct, it's a different tone being sung by the same species. One approach, as you suggest, could be to predict the song, not the species, and <strong>then</strong> use that information to help predict what species it is.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1103088,
          "author_name": "hazarre",
          "author_url": "",
          "post_date": "12/05/2020 16:12:18",
          "content": "<p>Awesome. Thank you.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1122107,
          "author_name": "swaralipibose",
          "author_url": "",
          "post_date": "12/22/2020 07:38:52",
          "content": "<p>also in submission.csv file they are one hot encoded 1 -&gt; s1 2-&gt;s2</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1102466": "Hi, \n\nI don't understand what to make of the songtype feature in the dataset? I am guess that it describes the different tones made by the same species. However, in that case, it seems that songtype should be taken as label instead of data, so should our model also learn songtype (not required by the contest) as well ?",
    "1102570": "You're correct, it's a different tone being sung by the same species. One approach, as you suggest, could be to predict the song, not the species, and **then** use that information to help predict what species it is.",
    "1103088": "Awesome. Thank you.",
    "1122107": "also in submission.csv file they are one hot encoded 1 -> s1 2->s2"
  },
  "source": "meta"
}