{
  "id": 94689,
  "title": "Custom LGBM Metric",
  "url": "/competitions/champs-scalar-coupling/discussion/94689",
  "author_name": "",
  "post_date": "2019-06-06T08:37:10.707880400Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>@uberkinder posted a great kernel demonstrating how to calculate the evaluation metric for this competition <a href=\"https://www.kaggle.com/uberkinder/efficient-metric\">here</a>.</p>\n\n<p>As a noob, I was wondering how this could be used as a LightGBM custom evaluation metric? I'm comfortable with how to do create the calculation, but not sure how to loop over the groups?</p>\n\n<p>Any hints / ideas would be great!</p>",
  "messages": [
    {
      "id": "546085",
      "postDate": "06/06/2019 08:37:10",
      "content": "<p>@uberkinder posted a great kernel demonstrating how to calculate the evaluation metric for this competition <a href=\"https://www.kaggle.com/uberkinder/efficient-metric\">here</a>.</p>\n\n<p>As a noob, I was wondering how this could be used as a LightGBM custom evaluation metric? I'm comfortable with how to do create the calculation, but not sure how to loop over the groups?</p>\n\n<p>Any hints / ideas would be great!</p>",
      "rawMarkdown": "uberkinder posted a great kernel demonstrating how to calculate the evaluation metric for this competition [here](https://www.kaggle.com/uberkinder/efficient-metric).\n\nAs a noob, I was wondering how this could be used as a LightGBM custom evaluation metric? I'm comfortable with how to do create the calculation, but not sure how to loop over the groups?\n\nAny hints / ideas would be great!",
      "votes": null
    },
    {
      "id": "546117",
      "postDate": "06/06/2019 09:14:48",
      "content": "<p>have a look at this kernel for reference:\n<a href=\"https://www.kaggle.com/artgor/brute-force-feature-engineering\">https://www.kaggle.com/artgor/brute-force-feature-engineering</a></p>",
      "rawMarkdown": "have a look at this kernel for reference:\nhttps://www.kaggle.com/artgor/brute-force-feature-engineering",
      "votes": null
    },
    {
      "id": "546175",
      "postDate": "06/06/2019 10:27:21",
      "content": "<p>Thank you!</p>",
      "rawMarkdown": "Thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 546117,
      "author_name": "something4kag",
      "author_url": "",
      "post_date": "06/06/2019 09:14:48",
      "content": "<p>have a look at this kernel for reference:\n<a href=\"https://www.kaggle.com/artgor/brute-force-feature-engineering\">https://www.kaggle.com/artgor/brute-force-feature-engineering</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 546175,
          "author_name": "bradsshaw",
          "author_url": "",
          "post_date": "06/06/2019 10:27:21",
          "content": "<p>Thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "546085": "uberkinder posted a great kernel demonstrating how to calculate the evaluation metric for this competition [here](https://www.kaggle.com/uberkinder/efficient-metric).\n\nAs a noob, I was wondering how this could be used as a LightGBM custom evaluation metric? I'm comfortable with how to do create the calculation, but not sure how to loop over the groups?\n\nAny hints / ideas would be great!",
    "546117": "have a look at this kernel for reference:\nhttps://www.kaggle.com/artgor/brute-force-feature-engineering",
    "546175": "Thank you!"
  },
  "source": "meta"
}