{
  "id": 132026,
  "title": "Confusion Matrix",
  "url": "/competitions/bengaliai-cv19/discussion/132026",
  "author_name": "",
  "post_date": "2020-02-23T13:55:52.960935Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hey, I train a separate model for Grapheme Root and the model achieves nearly 95% on training and validation accuracy, I predict some samples and plot their confusion matrix, and the confusion matrix is like <a href=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4439732%2F9c6128d2853a11d3ec62d8308f2c504d%2Fconfusion_matrix.html?generation=1582465774523465&amp;alt=media\">https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4439732%2F9c6128d2853a11d3ec62d8308f2c504d%2Fconfusion_matrix.html?generation=1582465774523465&amp;alt=media</a>. The confusion matrix looks something weird to me. Is this a normal confusion matrix?I will be very appreciated if anyone can help me 🙂</p>",
  "messages": [
    {
      "id": "754398",
      "postDate": "02/23/2020 13:55:52",
      "content": "<p>Hey, I train a separate model for Grapheme Root and the model achieves nearly 95% on training and validation accuracy, I predict some samples and plot their confusion matrix, and the confusion matrix is like <a href=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4439732%2F9c6128d2853a11d3ec62d8308f2c504d%2Fconfusion_matrix.html?generation=1582465774523465&amp;alt=media\">https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4439732%2F9c6128d2853a11d3ec62d8308f2c504d%2Fconfusion_matrix.html?generation=1582465774523465&amp;alt=media</a>. The confusion matrix looks something weird to me. Is this a normal confusion matrix?I will be very appreciated if anyone can help me 🙂</p>",
      "rawMarkdown": "Hey, I train a separate model for Grapheme Root and the model achieves nearly 95% on training and validation accuracy, I predict some samples and plot their confusion matrix, and the confusion matrix is like https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4439732%2F9c6128d2853a11d3ec62d8308f2c504d%2Fconfusion_matrix.html?generation=1582465774523465&amp;alt=media. The confusion matrix looks something weird to me. Is this a normal confusion matrix?I will be very appreciated if anyone can help me 🙂",
      "votes": null
    },
    {
      "id": "763456",
      "postDate": "03/04/2020 13:58:06",
      "content": "<p>Your confusion matrix is a color coded heatmap, rather than displaying raw numerical values.</p>\n\n<p>The \"confusion\" over the unusal looking natiure of it may be due to the fact that there are a very large number (168) of possible labels for your dataset.</p>\n\n<p>Else you would need to render a 168x168 matrix (it would be spreadsheet sizedm but it wouldn't easily fit on your screen)</p>",
      "rawMarkdown": "Your confusion matrix is a color coded heatmap, rather than displaying raw numerical values.\n\nThe \"confusion\" over the unusal looking natiure of it may be due to the fact that there are a very large number (168) of possible labels for your dataset.\n\nElse you would need to render a 168x168 matrix (it would be spreadsheet sizedm but it wouldn't easily fit on your screen)",
      "votes": null
    },
    {
      "id": "763548",
      "postDate": "03/04/2020 15:35:38",
      "content": "<p>Thanks for replying 🙂, but I solve the project before, it's because the model dataset was shuffled during testing, and that's why it show unusual behaviour of Confusion Matrix.</p>",
      "rawMarkdown": "Thanks for replying 🙂, but I solve the project before, it's because the model dataset was shuffled during testing, and that's why it show unusual behaviour of Confusion Matrix.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 763456,
      "author_name": "jamesmcguigan",
      "author_url": "",
      "post_date": "03/04/2020 13:58:06",
      "content": "<p>Your confusion matrix is a color coded heatmap, rather than displaying raw numerical values.</p>\n\n<p>The \"confusion\" over the unusal looking natiure of it may be due to the fact that there are a very large number (168) of possible labels for your dataset.</p>\n\n<p>Else you would need to render a 168x168 matrix (it would be spreadsheet sizedm but it wouldn't easily fit on your screen)</p>",
      "votes": null,
      "replies": [
        {
          "id": 763548,
          "author_name": "shubhamai",
          "author_url": "",
          "post_date": "03/04/2020 15:35:38",
          "content": "<p>Thanks for replying 🙂, but I solve the project before, it's because the model dataset was shuffled during testing, and that's why it show unusual behaviour of Confusion Matrix.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "754398": "Hey, I train a separate model for Grapheme Root and the model achieves nearly 95% on training and validation accuracy, I predict some samples and plot their confusion matrix, and the confusion matrix is like https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4439732%2F9c6128d2853a11d3ec62d8308f2c504d%2Fconfusion_matrix.html?generation=1582465774523465&amp;alt=media. The confusion matrix looks something weird to me. Is this a normal confusion matrix?I will be very appreciated if anyone can help me 🙂",
    "763456": "Your confusion matrix is a color coded heatmap, rather than displaying raw numerical values.\n\nThe \"confusion\" over the unusal looking natiure of it may be due to the fact that there are a very large number (168) of possible labels for your dataset.\n\nElse you would need to render a 168x168 matrix (it would be spreadsheet sizedm but it wouldn't easily fit on your screen)",
    "763548": "Thanks for replying 🙂, but I solve the project before, it's because the model dataset was shuffled during testing, and that's why it show unusual behaviour of Confusion Matrix."
  },
  "source": "meta"
}