{
  "id": 207195,
  "title": "How to use ranking over multi label? ",
  "url": "/competitions/ranzcr-clip-catheter-line-classification/discussion/207195",
  "author_name": "",
  "post_date": "2020-12-28T16:10:35.373438600Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi, I'm new to ML and kaggle.  Can anyone show me how to do ranking over the multi label prediction. As we have <code>auc</code> as metrics, I think it's necessary. </p>\n<pre><code>from scipy.stats import rankdata\n\nrankdata(df[target_cols].astype(float).values)\n</code></pre>",
  "messages": [
    {
      "id": "1129895",
      "postDate": "12/28/2020 16:10:35",
      "content": "<p>Hi, I'm new to ML and kaggle.  Can anyone show me how to do ranking over the multi label prediction. As we have <code>auc</code> as metrics, I think it's necessary. </p>\n<pre><code>from scipy.stats import rankdata\n\nrankdata(df[target_cols].astype(float).values)\n</code></pre>",
      "rawMarkdown": "Hi, I'm new to ML and kaggle.  Can anyone show me how to do ranking over the multi label prediction. As we have `auc` as metrics, I think it's necessary. \n\n```\nfrom scipy.stats import rankdata\n\nrankdata(df[target_cols].astype(float).values)\n```",
      "votes": null
    },
    {
      "id": "1130134",
      "postDate": "12/28/2020 18:56:39",
      "content": "<p>This is what you need:</p>\n<p><code>df[target_cols].apply(rankdata,axis=0).values</code></p>",
      "rawMarkdown": "This is what you need:\n\n`df[target_cols].apply(rankdata,axis=0).values`",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1130134,
      "author_name": "raddar",
      "author_url": "",
      "post_date": "12/28/2020 18:56:39",
      "content": "<p>This is what you need:</p>\n<p><code>df[target_cols].apply(rankdata,axis=0).values</code></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1129895": "Hi, I'm new to ML and kaggle.  Can anyone show me how to do ranking over the multi label prediction. As we have `auc` as metrics, I think it's necessary. \n\n```\nfrom scipy.stats import rankdata\n\nrankdata(df[target_cols].astype(float).values)\n```",
    "1130134": "This is what you need:\n\n`df[target_cols].apply(rankdata,axis=0).values`"
  },
  "source": "meta"
}