{
  "id": 232766,
  "title": "Number of Labels to Classify",
  "url": "/competitions/plant-pathology-2021-fgvc8/discussion/232766",
  "author_name": "",
  "post_date": "2021-04-15T09:35:48.150802400Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Could someone please clarify to me how many labels are actually there?<br>\n<strong>powdery_mildew complex</strong> would be treated as a single label or 2 labels **powdery_mildew ** and <strong>complex</strong>?</p>\n<p>Its the latter then it would be a multi-label problem</p>",
  "messages": [
    {
      "id": "1274420",
      "postDate": "04/15/2021 09:35:48",
      "content": "<p>Could someone please clarify to me how many labels are actually there?<br>\n<strong>powdery_mildew complex</strong> would be treated as a single label or 2 labels **powdery_mildew ** and <strong>complex</strong>?</p>\n<p>Its the latter then it would be a multi-label problem</p>",
      "rawMarkdown": "Could someone please clarify to me how many labels are actually there?\n**powdery_mildew complex** would be treated as a single label or 2 labels **powdery_mildew ** and **complex**?\n\nIts the latter then it would be a multi-label problem",
      "votes": null
    },
    {
      "id": "1275124",
      "postDate": "04/16/2021 01:02:28",
      "content": "<p>Hi there,</p>\n<p>In Exploratory Data Analysis using just 'train.csv' file you can check that there are 12 classes. Plot bar chart or pie with matplotlib.</p>\n<p>Then, 'powdery_mildew', 'complex' and  'powdery_mildew complex' are 3 different classes. Probably, 'powdery_mildew complex' has features from the 2 other single conditions combined.</p>\n<p>To confirm that, you can prepare a notebook that output a file with only one condition. </p>\n<p>For example, if you classify all the images as ''powdery_mildew', you get 0.063 in public score. On the other hand, 'powdery_mildew complex' gets 0.112</p>\n<p>I hope it helps.</p>",
      "rawMarkdown": "Hi there,\n\nIn Exploratory Data Analysis using just 'train.csv' file you can check that there are 12 classes. Plot bar chart or pie with matplotlib.\n\nThen, 'powdery_mildew', 'complex' and  'powdery_mildew complex' are 3 different classes. Probably, 'powdery_mildew complex' has features from the 2 other single conditions combined.\n\nTo confirm that, you can prepare a notebook that output a file with only one condition. \n\nFor example, if you classify all the images as ''powdery_mildew', you get 0.063 in public score. On the other hand, 'powdery_mildew complex' gets 0.112\n\nI hope it helps.",
      "votes": null
    },
    {
      "id": "1275544",
      "postDate": "04/16/2021 12:19:51",
      "content": "<p>This is a multilabel problem, see <a href=\"https://www.kaggle.com/c/plant-pathology-2021-fgvc8/discussion/226327#1240899\" target=\"_blank\">this reply from the staff member</a>.</p>",
      "rawMarkdown": "This is a multilabel problem, see [this reply from the staff member](https://www.kaggle.com/c/plant-pathology-2021-fgvc8/discussion/226327#1240899).",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1275124,
      "author_name": "thalesgaluchi",
      "author_url": "",
      "post_date": "04/16/2021 01:02:28",
      "content": "<p>Hi there,</p>\n<p>In Exploratory Data Analysis using just 'train.csv' file you can check that there are 12 classes. Plot bar chart or pie with matplotlib.</p>\n<p>Then, 'powdery_mildew', 'complex' and  'powdery_mildew complex' are 3 different classes. Probably, 'powdery_mildew complex' has features from the 2 other single conditions combined.</p>\n<p>To confirm that, you can prepare a notebook that output a file with only one condition. </p>\n<p>For example, if you classify all the images as ''powdery_mildew', you get 0.063 in public score. On the other hand, 'powdery_mildew complex' gets 0.112</p>\n<p>I hope it helps.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1275544,
      "author_name": "atamazian",
      "author_url": "",
      "post_date": "04/16/2021 12:19:51",
      "content": "<p>This is a multilabel problem, see <a href=\"https://www.kaggle.com/c/plant-pathology-2021-fgvc8/discussion/226327#1240899\" target=\"_blank\">this reply from the staff member</a>.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1274420": "Could someone please clarify to me how many labels are actually there?\n**powdery_mildew complex** would be treated as a single label or 2 labels **powdery_mildew ** and **complex**?\n\nIts the latter then it would be a multi-label problem",
    "1275124": "Hi there,\n\nIn Exploratory Data Analysis using just 'train.csv' file you can check that there are 12 classes. Plot bar chart or pie with matplotlib.\n\nThen, 'powdery_mildew', 'complex' and  'powdery_mildew complex' are 3 different classes. Probably, 'powdery_mildew complex' has features from the 2 other single conditions combined.\n\nTo confirm that, you can prepare a notebook that output a file with only one condition. \n\nFor example, if you classify all the images as ''powdery_mildew', you get 0.063 in public score. On the other hand, 'powdery_mildew complex' gets 0.112\n\nI hope it helps.",
    "1275544": "This is a multilabel problem, see [this reply from the staff member](https://www.kaggle.com/c/plant-pathology-2021-fgvc8/discussion/226327#1240899)."
  },
  "source": "meta"
}