{
  "id": 412366,
  "title": "why i (fold index) is not considered when saving the models?",
  "url": "/competitions/predict-student-performance-from-game-play/discussion/412366",
  "author_name": "yoshitown",
  "post_date": "2023-05-23T12:15:22.182000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi,everyone<br>\nwhy i (fold index) is not considered when saving the models?(I don't think that's  a mistake. I try to infer the meaning )<br>\nIn this code(<a href=\"https://www.kaggle.com/code/cdeotte/xgboost-baseline-0-680/comments)\" target=\"_blank\">https://www.kaggle.com/code/cdeotte/xgboost-baseline-0-680/comments)</a>, the models are saved in a dictionary called models using the keys '{grp}{t}' and the corresponding trained classifiers (RandomForestClassifier) as values. The key is formed by combining grp and '' (underscore), and t takes values from 1 to 18. This key uniquely identifies the model based on each question ('q') and group (grp)<br>\nI think the same model architecture and parameters are used for each fold's training, i is not considered when saving the models. The models are trained independently for each fold and are used during the final prediction step.but models are not same when models have different train_data. even if models are he same model architecture.<br>\nwhy i (fold index) is not considered when saving the models? please tell me your answer. </p>",
  "messages": [
    {
      "id": 2270784,
      "postDate": "2023-05-23T12:15:22.183Z",
      "content": "<p>Hi,everyone<br>\nwhy i (fold index) is not considered when saving the models?(I don't think that's  a mistake. I try to infer the meaning )<br>\nIn this code(<a href=\"https://www.kaggle.com/code/cdeotte/xgboost-baseline-0-680/comments)\" target=\"_blank\">https://www.kaggle.com/code/cdeotte/xgboost-baseline-0-680/comments)</a>, the models are saved in a dictionary called models using the keys '{grp}{t}' and the corresponding trained classifiers (RandomForestClassifier) as values. The key is formed by combining grp and '' (underscore), and t takes values from 1 to 18. This key uniquely identifies the model based on each question ('q') and group (grp)<br>\nI think the same model architecture and parameters are used for each fold's training, i is not considered when saving the models. The models are trained independently for each fold and are used during the final prediction step.but models are not same when models have different train_data. even if models are he same model architecture.<br>\nwhy i (fold index) is not considered when saving the models? please tell me your answer. </p>",
      "rawMarkdown": "Hi,everyone\nwhy i (fold index) is not considered when saving the models?(I don't think that's  a mistake. I try to infer the meaning )\nIn this code(https://www.kaggle.com/code/cdeotte/xgboost-baseline-0-680/comments), the models are saved in a dictionary called models using the keys '{grp}{t}' and the corresponding trained classifiers (RandomForestClassifier) as values. The key is formed by combining grp and '' (underscore), and t takes values from 1 to 18. This key uniquely identifies the model based on each question ('q') and group (grp)\nI think the same model architecture and parameters are used for each fold's training, i is not considered when saving the models. The models are trained independently for each fold and are used during the final prediction step.but models are not same when models have different train_data. even if models are he same model architecture.\nwhy i (fold index) is not considered when saving the models? please tell me your answer. "
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2270784": "Hi,everyone\nwhy i (fold index) is not considered when saving the models?(I don't think that's  a mistake. I try to infer the meaning )\nIn this code(https://www.kaggle.com/code/cdeotte/xgboost-baseline-0-680/comments), the models are saved in a dictionary called models using the keys '{grp}{t}' and the corresponding trained classifiers (RandomForestClassifier) as values. The key is formed by combining grp and '' (underscore), and t takes values from 1 to 18. This key uniquely identifies the model based on each question ('q') and group (grp)\nI think the same model architecture and parameters are used for each fold's training, i is not considered when saving the models. The models are trained independently for each fold and are used during the final prediction step.but models are not same when models have different train_data. even if models are he same model architecture.\nwhy i (fold index) is not considered when saving the models? please tell me your answer. "
  }
}