{
  "id": 395920,
  "title": "Why we can't use decision tree to solve this problem?",
  "url": "/competitions/predict-student-performance-from-game-play/discussion/395920",
  "author_name": "",
  "post_date": "2023-03-19T14:33:23.705049600Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi bosses I'm a y1 CS student, the only thing uni taught us so far is decision tree for binary classification problems. Why we can't use this model for this problem? What's the difference between this model and all the other fancy models you're using? </p>\n<p>Thank you guys and all your kind responses will be greatly appreciated.</p>",
  "messages": [
    {
      "id": "2188352",
      "postDate": "03/19/2023 14:33:23",
      "content": "<p>Hi bosses I'm a y1 CS student, the only thing uni taught us so far is decision tree for binary classification problems. Why we can't use this model for this problem? What's the difference between this model and all the other fancy models you're using? </p>\n<p>Thank you guys and all your kind responses will be greatly appreciated.</p>",
      "rawMarkdown": "Hi bosses I'm a y1 CS student, the only thing uni taught us so far is decision tree for binary classification problems. Why we can't use this model for this problem? What's the difference between this model and all the other fancy models you're using? \n\nThank you guys and all your kind responses will be greatly appreciated.",
      "votes": null
    },
    {
      "id": "2188386",
      "postDate": "03/19/2023 15:06:38",
      "content": "<p>You can use decision tree model too if you want for this competition or any other model which is capable of solving binary classification but a single decision tree often overfits on training data(we get a good score on train data but poor score on test data) so we use other more complicated models like xgboost, catboost, randomforest which basically uses many smaller decision trees and generalize the model better. Besides you can use gpu train with these fancy models which makes training faster</p>",
      "rawMarkdown": "You can use decision tree model too if you want for this competition or any other model which is capable of solving binary classification but a single decision tree often overfits on training data(we get a good score on train data but poor score on test data) so we use other more complicated models like xgboost, catboost, randomforest which basically uses many smaller decision trees and generalize the model better. Besides you can use gpu train with these fancy models which makes training faster",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2188386,
      "author_name": "shoabahamed",
      "author_url": "",
      "post_date": "03/19/2023 15:06:38",
      "content": "<p>You can use decision tree model too if you want for this competition or any other model which is capable of solving binary classification but a single decision tree often overfits on training data(we get a good score on train data but poor score on test data) so we use other more complicated models like xgboost, catboost, randomforest which basically uses many smaller decision trees and generalize the model better. Besides you can use gpu train with these fancy models which makes training faster</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2188352": "Hi bosses I'm a y1 CS student, the only thing uni taught us so far is decision tree for binary classification problems. Why we can't use this model for this problem? What's the difference between this model and all the other fancy models you're using? \n\nThank you guys and all your kind responses will be greatly appreciated.",
    "2188386": "You can use decision tree model too if you want for this competition or any other model which is capable of solving binary classification but a single decision tree often overfits on training data(we get a good score on train data but poor score on test data) so we use other more complicated models like xgboost, catboost, randomforest which basically uses many smaller decision trees and generalize the model better. Besides you can use gpu train with these fancy models which makes training faster"
  },
  "source": "meta"
}