Firstly, I have successful submission quite a few times. But, I don't know the reason this time. I made a pipeline function to process data before prediction. I tried this pipeline function for my test data and there's no issue. All predictions have values. I also double checked all features which should have values after the pipeline processing function. I keep receiving submission scoring errors for 6 times.
", "messages": [ { "id": "1137125", "postDate": "01/03/2021 17:44:21", "content": "Firstly, I have successful submission quite a few times. But, I don't know the reason this time. I made a pipeline function to process data before prediction. I tried this pipeline function for my test data and there's no issue. All predictions have values. I also double checked all features which should have values after the pipeline processing function. I keep receiving submission scoring errors for 6 times.
", "rawMarkdown": "Firstly, I have successful submission quite a few times. But, I don't know the reason this time. I made a pipeline function to process data before prediction. I tried this pipeline function for my test data and there's no issue. All predictions have values. I also double checked all features which should have values after the pipeline processing function. I keep receiving submission scoring errors for 6 times.", "votes": null }, { "id": "1137132", "postDate": "01/03/2021 17:46:50", "content": "The error shows at 27 mins at the latest try, which I just used 1,000,000 data for a quick try.
", "rawMarkdown": "The error shows at 27 mins at the latest try, which I just used 1,000,000 data for a quick try.", "votes": null }, { "id": "1137171", "postDate": "01/03/2021 18:25:58", "content": "From my experience, assuming you pass the local test:
\nif the error shows in a few minutes after submission, there are NaN values that you do not process when you inference.
\nif the error shows after 9 hrs, you may run out of memory or run out of runtime.
Submission with errors also counts in 5 submission per day?! This is so frustrated!! I made the feature very simple, only two. But still has submission error. And I cannot check exactly where's wrong!!
", "rawMarkdown": "Submission with errors also counts in 5 submission per day?! This is so frustrated!! I made the feature very simple, only two. But still has submission error. And I cannot check exactly where's wrong!!", "votes": null }, { "id": "1137302", "postDate": "01/03/2021 20:20:00", "content": "You know what? I made a new test file which contains new users and new contents. The prediction can be done without any problems. There's no NaN for any prediction probability. This is also what I expected because I've made sure all features will have values during the prediction. But WHY there's still scoring errors!!!
\nCan Anyone managing the system check?
", "rawMarkdown": "You know what? I made a new test file which contains new users and new contents. The prediction can be done without any problems. There's no NaN for any prediction probability. This is also what I expected because I've made sure all features will have values during the prediction. But WHY there's still scoring errors!!! \n\nCan Anyone managing the system check?", "votes": null } ], "comments": [ { "id": 1137132, "author_name": "noahxi", "author_url": "", "post_date": "01/03/2021 17:46:50", "content": "The error shows at 27 mins at the latest try, which I just used 1,000,000 data for a quick try.
", "votes": null, "replies": [] }, { "id": 1137171, "author_name": "woshifym", "author_url": "", "post_date": "01/03/2021 18:25:58", "content": "From my experience, assuming you pass the local test:
\nif the error shows in a few minutes after submission, there are NaN values that you do not process when you inference.
\nif the error shows after 9 hrs, you may run out of memory or run out of runtime.
Submission with errors also counts in 5 submission per day?! This is so frustrated!! I made the feature very simple, only two. But still has submission error. And I cannot check exactly where's wrong!!
", "votes": null, "replies": [] }, { "id": 1137302, "author_name": "noahxi", "author_url": "", "post_date": "01/03/2021 20:20:00", "content": "You know what? I made a new test file which contains new users and new contents. The prediction can be done without any problems. There's no NaN for any prediction probability. This is also what I expected because I've made sure all features will have values during the prediction. But WHY there's still scoring errors!!!
\nCan Anyone managing the system check?
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1137125": "Firstly, I have successful submission quite a few times. But, I don't know the reason this time. I made a pipeline function to process data before prediction. I tried this pipeline function for my test data and there's no issue. All predictions have values. I also double checked all features which should have values after the pipeline processing function. I keep receiving submission scoring errors for 6 times.", "1137132": "The error shows at 27 mins at the latest try, which I just used 1,000,000 data for a quick try.", "1137171": "From my experience, assuming you pass the local test:\nif the error shows in a few minutes after submission, there are NaN values that you do not process when you inference.\nif the error shows after 9 hrs, you may run out of memory or run out of runtime.", "1137276": "Submission with errors also counts in 5 submission per day?! This is so frustrated!! I made the feature very simple, only two. But still has submission error. And I cannot check exactly where's wrong!!", "1137302": "You know what? I made a new test file which contains new users and new contents. The prediction can be done without any problems. There's no NaN for any prediction probability. This is also what I expected because I've made sure all features will have values during the prediction. But WHY there's still scoring errors!!! \n\nCan Anyone managing the system check?" }, "source": "meta" }