{
  "id": 199956,
  "title": "Data imbalanced ",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/199956",
  "author_name": "",
  "post_date": "2020-11-28T05:24:27.989542500Z",
  "votes": 2,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Since there are 10,000 images for label 3 and around 2k for others almost. Will we consider this as data imbalanced? </p>",
  "messages": [
    {
      "id": "1093891",
      "postDate": "11/28/2020 05:24:27",
      "content": "<p>Since there are 10,000 images for label 3 and around 2k for others almost. Will we consider this as data imbalanced? </p>",
      "rawMarkdown": "Since there are 10,000 images for label 3 and around 2k for others almost. Will we consider this as data imbalanced?",
      "votes": null
    },
    {
      "id": "1094435",
      "postDate": "11/28/2020 16:03:48",
      "content": "<p>Well, I would say that this is not the most heavily unbalanced dataset I have seen. Normally, if you look at the dataset used in Fraud Detection, or even to diagnose a disease, obviously they're far more unbalanced and the positive class, the one we're interested to predict, has few examples compared to other classes.<br>\nIn this case, there is some unbalance.</p>\n<p>Probably, your question is: is it worth the effort to try some of the methods commonly used to handle class imbalance?</p>",
      "rawMarkdown": "Well, I would say that this is not the most heavily unbalanced dataset I have seen. Normally, if you look at the dataset used in Fraud Detection, or even to diagnose a disease, obviously they're far more unbalanced and the positive class, the one we're interested to predict, has few examples compared to other classes.\nIn this case, there is some unbalance.\n\nProbably, your question is: is it worth the effort to try some of the methods commonly used to handle class imbalance?",
      "votes": null
    },
    {
      "id": "1094437",
      "postDate": "11/28/2020 16:04:30",
      "content": "<p>Ah, I have not adopted any of these methods, so far.</p>",
      "rawMarkdown": "Ah, I have not adopted any of these methods, so far.",
      "votes": null
    },
    {
      "id": "1094452",
      "postDate": "11/28/2020 16:16:56",
      "content": "<p>yeah, on point.</p>",
      "rawMarkdown": "yeah, on point.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1094435,
      "author_name": "luigisaetta",
      "author_url": "",
      "post_date": "11/28/2020 16:03:48",
      "content": "<p>Well, I would say that this is not the most heavily unbalanced dataset I have seen. Normally, if you look at the dataset used in Fraud Detection, or even to diagnose a disease, obviously they're far more unbalanced and the positive class, the one we're interested to predict, has few examples compared to other classes.<br>\nIn this case, there is some unbalance.</p>\n<p>Probably, your question is: is it worth the effort to try some of the methods commonly used to handle class imbalance?</p>",
      "votes": null,
      "replies": [
        {
          "id": 1094452,
          "author_name": "mtalhaarshad",
          "author_url": "",
          "post_date": "11/28/2020 16:16:56",
          "content": "<p>yeah, on point.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1094437,
      "author_name": "luigisaetta",
      "author_url": "",
      "post_date": "11/28/2020 16:04:30",
      "content": "<p>Ah, I have not adopted any of these methods, so far.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1093891": "Since there are 10,000 images for label 3 and around 2k for others almost. Will we consider this as data imbalanced?",
    "1094435": "Well, I would say that this is not the most heavily unbalanced dataset I have seen. Normally, if you look at the dataset used in Fraud Detection, or even to diagnose a disease, obviously they're far more unbalanced and the positive class, the one we're interested to predict, has few examples compared to other classes.\nIn this case, there is some unbalance.\n\nProbably, your question is: is it worth the effort to try some of the methods commonly used to handle class imbalance?",
    "1094437": "Ah, I have not adopted any of these methods, so far.",
    "1094452": "yeah, on point."
  },
  "source": "meta"
}