{
  "id": 400535,
  "title": "Whats the best metric to tune against?",
  "url": "/competitions/predict-student-performance-from-game-play/discussion/400535",
  "author_name": "",
  "post_date": "2023-04-08T22:08:21.676722800Z",
  "votes": null,
  "comment_count": 5,
  "views": 0,
  "content": "<p>I am currently tuning against log loss.</p>\n<p>What is everyone one else tuning against?</p>\n<p>What is the best one?</p>",
  "messages": [
    {
      "id": "2214934",
      "postDate": "04/08/2023 22:08:21",
      "content": "<p>I am currently tuning against log loss.</p>\n<p>What is everyone one else tuning against?</p>\n<p>What is the best one?</p>",
      "rawMarkdown": "I am currently tuning against log loss.\n\nWhat is everyone one else tuning against?\n\nWhat is the best one?",
      "votes": null
    },
    {
      "id": "2215005",
      "postDate": "04/09/2023 01:51:03",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/gordyii\" target=\"_blank\">@gordyii</a>  In fact, there is no one-size-fits-all metric for tuning a model as it largely depends on the nature of the dataset and the specific model being used. However, when considering the evaluation of this submissions, the F1 score might be a suitable metric to begin with. </p>",
      "rawMarkdown": "Hi @gordyii  In fact, there is no one-size-fits-all metric for tuning a model as it largely depends on the nature of the dataset and the specific model being used. However, when considering the evaluation of this submissions, the F1 score might be a suitable metric to begin with.",
      "votes": null
    },
    {
      "id": "2216664",
      "postDate": "04/10/2023 08:54:05",
      "content": "<p>Thanks very much for the information :) </p>",
      "rawMarkdown": "Thanks very much for the information :)",
      "votes": null
    },
    {
      "id": "2218680",
      "postDate": "04/11/2023 23:16:32",
      "content": "<p>Which eval_metric is best for XGBoost for this problem?  There is no F1 there….?!?<br>\n<a href=\"https://xgboost.readthedocs.io/en/stable/parameter.html\" target=\"_blank\">https://xgboost.readthedocs.io/en/stable/parameter.html</a></p>",
      "rawMarkdown": "Which eval_metric is best for XGBoost for this problem?  There is no F1 there....?!?\nhttps://xgboost.readthedocs.io/en/stable/parameter.html",
      "votes": null
    },
    {
      "id": "2218687",
      "postDate": "04/11/2023 23:27:26",
      "content": "<p>You can customize the eval function or even use sklearn metrics.</p>\n<p><a href=\"https://xgboost.readthedocs.io/en/stable/tutorials/custom_metric_obj.html\" target=\"_blank\">https://xgboost.readthedocs.io/en/stable/tutorials/custom_metric_obj.html</a></p>",
      "rawMarkdown": "You can customize the eval function or even use sklearn metrics.\n\nhttps://xgboost.readthedocs.io/en/stable/tutorials/custom_metric_obj.html",
      "votes": null
    },
    {
      "id": "2218689",
      "postDate": "04/11/2023 23:30:21",
      "content": "<p>This <a href=\"https://stackoverflow.com/questions/51587535/custom-evaluation-function-based-on-f1-for-use-in-xgboost-python-api\" target=\"_blank\">Stackoverflow</a> had your F1 eval function, hope this helps </p>",
      "rawMarkdown": "This [Stackoverflow](https://stackoverflow.com/questions/51587535/custom-evaluation-function-based-on-f1-for-use-in-xgboost-python-api) had your F1 eval function, hope this helps",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2215005,
      "author_name": "thaweewatboy",
      "author_url": "",
      "post_date": "04/09/2023 01:51:03",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/gordyii\" target=\"_blank\">@gordyii</a>  In fact, there is no one-size-fits-all metric for tuning a model as it largely depends on the nature of the dataset and the specific model being used. However, when considering the evaluation of this submissions, the F1 score might be a suitable metric to begin with. </p>",
      "votes": null,
      "replies": [
        {
          "id": 2216664,
          "author_name": "gordyii",
          "author_url": "",
          "post_date": "04/10/2023 08:54:05",
          "content": "<p>Thanks very much for the information :) </p>",
          "votes": null,
          "replies": [
            {
              "id": 2218680,
              "author_name": "gordyii",
              "author_url": "",
              "post_date": "04/11/2023 23:16:32",
              "content": "<p>Which eval_metric is best for XGBoost for this problem?  There is no F1 there….?!?<br>\n<a href=\"https://xgboost.readthedocs.io/en/stable/parameter.html\" target=\"_blank\">https://xgboost.readthedocs.io/en/stable/parameter.html</a></p>",
              "votes": null,
              "replies": [
                {
                  "id": 2218687,
                  "author_name": "thaweewatboy",
                  "author_url": "",
                  "post_date": "04/11/2023 23:27:26",
                  "content": "<p>You can customize the eval function or even use sklearn metrics.</p>\n<p><a href=\"https://xgboost.readthedocs.io/en/stable/tutorials/custom_metric_obj.html\" target=\"_blank\">https://xgboost.readthedocs.io/en/stable/tutorials/custom_metric_obj.html</a></p>",
                  "votes": null,
                  "replies": []
                },
                {
                  "id": 2218689,
                  "author_name": "thaweewatboy",
                  "author_url": "",
                  "post_date": "04/11/2023 23:30:21",
                  "content": "<p>This <a href=\"https://stackoverflow.com/questions/51587535/custom-evaluation-function-based-on-f1-for-use-in-xgboost-python-api\" target=\"_blank\">Stackoverflow</a> had your F1 eval function, hope this helps </p>",
                  "votes": null,
                  "replies": []
                }
              ]
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2214934": "I am currently tuning against log loss.\n\nWhat is everyone one else tuning against?\n\nWhat is the best one?",
    "2215005": "Hi @gordyii  In fact, there is no one-size-fits-all metric for tuning a model as it largely depends on the nature of the dataset and the specific model being used. However, when considering the evaluation of this submissions, the F1 score might be a suitable metric to begin with.",
    "2216664": "Thanks very much for the information :)",
    "2218680": "Which eval_metric is best for XGBoost for this problem?  There is no F1 there....?!?\nhttps://xgboost.readthedocs.io/en/stable/parameter.html",
    "2218687": "You can customize the eval function or even use sklearn metrics.\n\nhttps://xgboost.readthedocs.io/en/stable/tutorials/custom_metric_obj.html",
    "2218689": "This [Stackoverflow](https://stackoverflow.com/questions/51587535/custom-evaluation-function-based-on-f1-for-use-in-xgboost-python-api) had your F1 eval function, hope this helps"
  },
  "source": "meta"
}