How should the CSV file be written because submission always fails if we had there something else than zeros?
\n\nWe returned the sample_submission.csv file with pure zeros and we can achieve the score. However, if we upload a our results CSV file to the server, read it and write it in a notebook as done in the example case (by using sample_submission.csv file) we get Submission scoring error.
", "messages": [ { "id": "764681", "postDate": "03/05/2020 18:21:01", "content": "How should the CSV file be written because submission always fails if we had there something else than zeros?
\n\nWe returned the sample_submission.csv file with pure zeros and we can achieve the score. However, if we upload a our results CSV file to the server, read it and write it in a notebook as done in the example case (by using sample_submission.csv file) we get Submission scoring error.
", "rawMarkdown": "How should the CSV file be written because submission always fails if we had there something else than zeros?\n\nWe returned the sample_submission.csv file with pure zeros and we can achieve the score. However, if we upload a our results CSV file to the server, read it and write it in a notebook as done in the example case (by using sample_submission.csv file) we get Submission scoring error.", "votes": null }, { "id": "765925", "postDate": "03/07/2020 11:23:29", "content": "+1
", "rawMarkdown": "1", "votes": null }, { "id": "767515", "postDate": "03/09/2020 18:25:56", "content": "The public leaderboard is being scored on a fully withheld private test set. So if you are hardcoding the id's from the sample submission or trying to read a .CSV that you've hardcoded with specific ids, it will fail on submission, because the private test sets id's will not be part of that hardcoded submission. You must write your code to flexibly take any set of input test_videos and write a submission.csv dynamically.
Thank you for the information Julia. We just wanted to return a CSV file just to make sure we are writing the CSV file properly and to quickly see our results without running the massive code in the server. But actually hard coding worked by reading the sample_submission.csv and changing the values in it and then creating the final submission file. However, we still don't know did it work because we used pandas library (instead of csv library) or we created the file in the Kaggle server instead of our own PC (which uses Linux which could affect the encoding).
", "rawMarkdown": "Thank you for the information Julia. We just wanted to return a CSV file just to make sure we are writing the CSV file properly and to quickly see our results without running the massive code in the server. But actually hard coding worked by reading the sample_submission.csv and changing the values in it and then creating the final submission file. However, we still don't know did it work because we used pandas library (instead of csv library) or we created the file in the Kaggle server instead of our own PC (which uses Linux which could affect the encoding).", "votes": null } ], "comments": [ { "id": 765925, "author_name": "kajjjak", "author_url": "", "post_date": "03/07/2020 11:23:29", "content": "+1
", "votes": null, "replies": [] }, { "id": 767515, "author_name": "juliaelliott", "author_url": "", "post_date": "03/09/2020 18:25:56", "content": "The public leaderboard is being scored on a fully withheld private test set. So if you are hardcoding the id's from the sample submission or trying to read a .CSV that you've hardcoded with specific ids, it will fail on submission, because the private test sets id's will not be part of that hardcoded submission. You must write your code to flexibly take any set of input test_videos and write a submission.csv dynamically.
Thank you for the information Julia. We just wanted to return a CSV file just to make sure we are writing the CSV file properly and to quickly see our results without running the massive code in the server. But actually hard coding worked by reading the sample_submission.csv and changing the values in it and then creating the final submission file. However, we still don't know did it work because we used pandas library (instead of csv library) or we created the file in the Kaggle server instead of our own PC (which uses Linux which could affect the encoding).
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "764681": "How should the CSV file be written because submission always fails if we had there something else than zeros?\n\nWe returned the sample_submission.csv file with pure zeros and we can achieve the score. However, if we upload a our results CSV file to the server, read it and write it in a notebook as done in the example case (by using sample_submission.csv file) we get Submission scoring error.", "765925": "1", "767515": "The public leaderboard is being scored on a fully withheld private test set. So if you are hardcoding the id's from the sample submission or trying to read a .CSV that you've hardcoded with specific ids, it will fail on submission, because the private test sets id's will not be part of that hardcoded submission. You must write your code to flexibly take any set of input `test_videos` and write a `submission.csv` dynamically.", "767889": "Thank you for the information Julia. We just wanted to return a CSV file just to make sure we are writing the CSV file properly and to quickly see our results without running the massive code in the server. But actually hard coding worked by reading the sample_submission.csv and changing the values in it and then creating the final submission file. However, we still don't know did it work because we used pandas library (instead of csv library) or we created the file in the Kaggle server instead of our own PC (which uses Linux which could affect the encoding)." }, "source": "meta" }