{
  "id": 402970,
  "title": "When should I start to tune my hyperparameter?",
  "url": "/competitions/predict-student-performance-from-game-play/discussion/402970",
  "author_name": "",
  "post_date": "2023-04-20T13:49:11.612537300Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Greetings, Kagglers,</p>\n<p>I am participating in my first Kaggle competition and have a question as a newcomer: When is the appropriate time to start fine-tuning my hyperparameters, such as the parameters in XGBoost? Should I concentrate on feature engineering and selection instead? Or I can co-tune it while I am doing feature selections. I am simply curious about the typical process for hyperparameter tuning. Thank you!</p>",
  "messages": [
    {
      "id": "2228366",
      "postDate": "04/20/2023 13:49:11",
      "content": "<p>Greetings, Kagglers,</p>\n<p>I am participating in my first Kaggle competition and have a question as a newcomer: When is the appropriate time to start fine-tuning my hyperparameters, such as the parameters in XGBoost? Should I concentrate on feature engineering and selection instead? Or I can co-tune it while I am doing feature selections. I am simply curious about the typical process for hyperparameter tuning. Thank you!</p>",
      "rawMarkdown": "Greetings, Kagglers,\n\nI am participating in my first Kaggle competition and have a question as a newcomer: When is the appropriate time to start fine-tuning my hyperparameters, such as the parameters in XGBoost? Should I concentrate on feature engineering and selection instead? Or I can co-tune it while I am doing feature selections. I am simply curious about the typical process for hyperparameter tuning. Thank you!",
      "votes": null
    },
    {
      "id": "2228773",
      "postDate": "04/20/2023 20:00:52",
      "content": "<p>You could tune parameters as you progress in the competition, submitting your experiment results as your CV improves too. I may suggest you could do as below-</p>\n<ol>\n<li>Look for better features/ transforms</li>\n<li>Tune parameters of models (all/ selected)</li>\n<li>Submit experiment results based on CV score improvement</li>\n<li>Note the experiment results and proceed with changes as per your CV and leaderboard scores</li>\n</ol>\n<p>All the best <a href=\"https://www.kaggle.com/mrhantato\" target=\"_blank\">@mrhantato</a> </p>",
      "rawMarkdown": "You could tune parameters as you progress in the competition, submitting your experiment results as your CV improves too. I may suggest you could do as below-\n1. Look for better features/ transforms\n2. Tune parameters of models (all/ selected)\n3. Submit experiment results based on CV score improvement\n4. Note the experiment results and proceed with changes as per your CV and leaderboard scores\n\nAll the best @mrhantato",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2228773,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "04/20/2023 20:00:52",
      "content": "<p>You could tune parameters as you progress in the competition, submitting your experiment results as your CV improves too. I may suggest you could do as below-</p>\n<ol>\n<li>Look for better features/ transforms</li>\n<li>Tune parameters of models (all/ selected)</li>\n<li>Submit experiment results based on CV score improvement</li>\n<li>Note the experiment results and proceed with changes as per your CV and leaderboard scores</li>\n</ol>\n<p>All the best <a href=\"https://www.kaggle.com/mrhantato\" target=\"_blank\">@mrhantato</a> </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2228366": "Greetings, Kagglers,\n\nI am participating in my first Kaggle competition and have a question as a newcomer: When is the appropriate time to start fine-tuning my hyperparameters, such as the parameters in XGBoost? Should I concentrate on feature engineering and selection instead? Or I can co-tune it while I am doing feature selections. I am simply curious about the typical process for hyperparameter tuning. Thank you!",
    "2228773": "You could tune parameters as you progress in the competition, submitting your experiment results as your CV improves too. I may suggest you could do as below-\n1. Look for better features/ transforms\n2. Tune parameters of models (all/ selected)\n3. Submit experiment results based on CV score improvement\n4. Note the experiment results and proceed with changes as per your CV and leaderboard scores\n\nAll the best @mrhantato"
  },
  "source": "meta"
}