{
  "id": 506833,
  "title": "Why metric hacking work in some cases only?",
  "url": "/competitions/home-credit-credit-risk-model-stability/discussion/506833",
  "author_name": "",
  "post_date": "2024-05-23T11:59:04.615421500Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>When I tried to implement the metric hacking technique in my different approaches, it is giving worse results? It is working with only some models. Why? </p>",
  "messages": [
    {
      "id": "2830871",
      "postDate": "05/23/2024 11:59:04",
      "content": "<p>When I tried to implement the metric hacking technique in my different approaches, it is giving worse results? It is working with only some models. Why? </p>",
      "rawMarkdown": "When I tried to implement the metric hacking technique in my different approaches, it is giving worse results? It is working with only some models. Why?",
      "votes": null
    },
    {
      "id": "2832889",
      "postDate": "05/23/2024 23:43:39",
      "content": "<p>Because some models are not affected  significantly by the second term of the metric (decline with time slope).</p>\n<p>This discussion has an example of a model that would benefit from the hack:<br>\n<a href=\"https://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/505987\" target=\"_blank\">https://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/505987</a></p>",
      "rawMarkdown": "Because some models are not affected  significantly by the second term of the metric (decline with time slope).\n\nThis discussion has an example of a model that would benefit from the hack:\nhttps://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/505987",
      "votes": null
    },
    {
      "id": "2834652",
      "postDate": "05/24/2024 21:52:19",
      "content": "<p>Thank you for confirming. I was also thinking the same.</p>",
      "rawMarkdown": "Thank you for confirming. I was also thinking the same.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2832889,
      "author_name": "fadylabib",
      "author_url": "",
      "post_date": "05/23/2024 23:43:39",
      "content": "<p>Because some models are not affected  significantly by the second term of the metric (decline with time slope).</p>\n<p>This discussion has an example of a model that would benefit from the hack:<br>\n<a href=\"https://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/505987\" target=\"_blank\">https://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/505987</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 2834652,
          "author_name": "shubhamchauhan22222",
          "author_url": "",
          "post_date": "05/24/2024 21:52:19",
          "content": "<p>Thank you for confirming. I was also thinking the same.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2830871": "When I tried to implement the metric hacking technique in my different approaches, it is giving worse results? It is working with only some models. Why?",
    "2832889": "Because some models are not affected  significantly by the second term of the metric (decline with time slope).\n\nThis discussion has an example of a model that would benefit from the hack:\nhttps://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/505987",
    "2834652": "Thank you for confirming. I was also thinking the same."
  },
  "source": "meta"
}