There are some problems in the train data of this competition. This very helpful discussion by @abaojiang discusses these problems in great detail, but we were unsure about whether these are present in the test data also or not.
\nI tested this for the error in elapsed time column and for multiple games present in a single session using notebooks which I have made public. You can find them here:
\nThe way I created the code, both the notebooks submissions were to fail if the problems are present in the test data also. And, both of these notebooks do fail.
\nSo,
\nYou can modify the notebooks to test for other problems also.
\nI hope this will help!!
There are some problems in the train data of this competition. This very helpful discussion by @abaojiang discusses these problems in great detail, but we were unsure about whether these are present in the test data also or not.
\nI tested this for the error in elapsed time column and for multiple games present in a single session using notebooks which I have made public. You can find them here:
\nThe way I created the code, both the notebooks submissions were to fail if the problems are present in the test data also. And, both of these notebooks do fail.
\nSo,
\nYou can modify the notebooks to test for other problems also.
\nI hope this will help!!
This is very interesting ! May I ask how you found those issues ?
", "rawMarkdown": "This is very interesting ! May I ask how you found those issues ?", "votes": 1, "replies": [ { "id": 2187181, "postDate": "2023-03-18T13:32:22.107Z", "content": "Hii, by doing some EDA!! ;)
", "rawMarkdown": "Hii, by doing some EDA!! ;)" } ] }, { "id": 2194732, "postDate": "2023-03-24T05:52:27.690Z", "content": "It looks like this affects almost the entire dataset. What do you think about imputing the 'elapsed_time' column with the median values between the affected indices and the next correct value?
", "rawMarkdown": "It looks like this affects almost the entire dataset. What do you think about imputing the 'elapsed_time' column with the median values between the affected indices and the next correct value?" }, { "id": 2188049, "postDate": "2023-03-19T08:59:22.573Z", "content": "Good work 👍
", "rawMarkdown": "Good work 👍", "votes": 1, "isDeleted": true }, { "id": 2187868, "postDate": "2023-03-19T05:13:34.877Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 2187067, "author_name": "JEANMPIA", "author_url": "", "post_date": "2023-03-18T11:36:14.843000", "content": "This is very interesting ! May I ask how you found those issues ?
", "votes": 1, "replies": [ { "id": 2187181, "author_name": "Shashwat Raman", "author_url": "", "post_date": "2023-03-18T13:32:22.107000", "content": "Hii, by doing some EDA!! ;)
", "votes": 0, "replies": [] } ] }, { "id": 2194732, "author_name": "Peter Nieuwkerk", "author_url": "", "post_date": "2023-03-24T05:52:27.690000", "content": "It looks like this affects almost the entire dataset. What do you think about imputing the 'elapsed_time' column with the median values between the affected indices and the next correct value?
", "votes": 0, "replies": [] }, { "id": 2188049, "author_name": "", "author_url": "", "post_date": "2023-03-19T08:59:22.573000", "content": "Good work 👍
", "votes": 1, "replies": [] }, { "id": 2187868, "author_name": "", "author_url": "", "post_date": "2023-03-19T05:13:34.877000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2186958": "# Problems in the Train Data are also present in the Test Data!!!\n\nThere are some problems in the train data of this competition. [This](https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/395250) very helpful discussion by @abaojiang discusses these problems in great detail, but we were unsure about whether these are present in the test data also or not. \n\nI tested this for the error in elapsed time column and for multiple games present in a single session using notebooks which I have made public. You can find them here:\n#####Elapsed Time Column: [Link](https://www.kaggle.com/code/shashwatraman/test-data-error-check-elapsed-time?scriptVersionId=122547167)\n######Multiple Games: [Link](https://www.kaggle.com/code/shashwatraman/test-data-error-check-multiple-games?scriptVersionId=122547117)\n\nThe way I created the code, both the notebooks submissions were to fail if the problems are present in the test data also. And, both of these notebooks do **fail**.\n\nSo,\n###### Elapsed Time Column does not monotonically increase in the test data also!!\n###### Multiple Games within a single session are also present in the test data!! (Kaggle API does provide a leak into the future in the test data also) (Answer to [this](https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/388479#2149959) question by @cdeotte)\n\nYou can modify the notebooks to test for other problems also.\nI hope this will help!!", "2187067": "This is very interesting ! May I ask how you found those issues ?", "2194732": "It looks like this affects almost the entire dataset. What do you think about imputing the 'elapsed_time' column with the median values between the affected indices and the next correct value?", "2188049": "Good work 👍", "2187868": "" } }