I have uploaded the right file with two columns 'id' and 'label', checked that 'label' is int. Also made sure I substituted my predictions in the sample submission file and then downloaded it. No matter what I do, I'm getting submission score as 0.0000\nCan anyone please help me?
", "messages": [ { "id": "685161", "postDate": "12/01/2019 04:19:58", "content": "I have uploaded the right file with two columns 'id' and 'label', checked that 'label' is int. Also made sure I substituted my predictions in the sample submission file and then downloaded it. No matter what I do, I'm getting submission score as 0.0000\nCan anyone please help me?
", "rawMarkdown": "I have uploaded the right file with two columns 'id' and 'label', checked that 'label' is int. Also made sure I substituted my predictions in the sample submission file and then downloaded it. No matter what I do, I'm getting submission score as 0.0000\nCan anyone please help me?", "votes": null }, { "id": "685610", "postDate": "12/02/2019 01:58:41", "content": "Are you predictions for each class using zero indexing?
", "rawMarkdown": "Are you predictions for each class using zero indexing?", "votes": null }, { "id": "686155", "postDate": "12/02/2019 20:38:17", "content": "When you do a submission the whole of your code is run again. It might be that the id's for the hidden test set are different. I suggest you take a look at one of the example kernels for example this one to see how they handle the process of generating the submission with the correct id's that will generate a return with non zero score.
", "rawMarkdown": "When you do a submission the whole of your code is run again. It might be that the id's for the hidden test set are different. I suggest you take a look at one of the example kernels for example [this one](https://www.kaggle.com/datahobbit/grandmaster-solution-adapted-from-mnist) to see how they handle the process of generating the submission with the correct id's that will generate a return with non zero score.", "votes": null }, { "id": "686757", "postDate": "12/03/2019 13:53:13", "content": "hi, if you scroll down your committed notebook, do you see top of your submission.csv file? if so, what do you see in first couple of rows in column 'label'? if else than 3 0 2 6 7, your prediction did not go well. also it may worth checking number of rows in your submission.csv file.
", "rawMarkdown": "hi, if you scroll down your committed notebook, do you see top of your submission.csv file? if so, what do you see in first couple of rows in column 'label'? if else than 3 0 2 6 7, your prediction did not go well. also it may worth checking number of rows in your submission.csv file.", "votes": null }, { "id": "687295", "postDate": "12/04/2019 07:33:45", "content": "start with a public kernel and only modify the predictions based on your algorithm.
", "rawMarkdown": "start with a public kernel and only modify the predictions based on your algorithm.", "votes": null } ], "comments": [ { "id": 685610, "author_name": "nirvaux", "author_url": "", "post_date": "12/02/2019 01:58:41", "content": "Are you predictions for each class using zero indexing?
", "votes": null, "replies": [] }, { "id": 686155, "author_name": "datahobbit", "author_url": "", "post_date": "12/02/2019 20:38:17", "content": "When you do a submission the whole of your code is run again. It might be that the id's for the hidden test set are different. I suggest you take a look at one of the example kernels for example this one to see how they handle the process of generating the submission with the correct id's that will generate a return with non zero score.
", "votes": null, "replies": [] }, { "id": 686757, "author_name": "nandor65", "author_url": "", "post_date": "12/03/2019 13:53:13", "content": "hi, if you scroll down your committed notebook, do you see top of your submission.csv file? if so, what do you see in first couple of rows in column 'label'? if else than 3 0 2 6 7, your prediction did not go well. also it may worth checking number of rows in your submission.csv file.
", "votes": null, "replies": [] }, { "id": 687295, "author_name": "rincewind007", "author_url": "", "post_date": "12/04/2019 07:33:45", "content": "start with a public kernel and only modify the predictions based on your algorithm.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "685161": "I have uploaded the right file with two columns 'id' and 'label', checked that 'label' is int. Also made sure I substituted my predictions in the sample submission file and then downloaded it. No matter what I do, I'm getting submission score as 0.0000\nCan anyone please help me?", "685610": "Are you predictions for each class using zero indexing?", "686155": "When you do a submission the whole of your code is run again. It might be that the id's for the hidden test set are different. I suggest you take a look at one of the example kernels for example [this one](https://www.kaggle.com/datahobbit/grandmaster-solution-adapted-from-mnist) to see how they handle the process of generating the submission with the correct id's that will generate a return with non zero score.", "686757": "hi, if you scroll down your committed notebook, do you see top of your submission.csv file? if so, what do you see in first couple of rows in column 'label'? if else than 3 0 2 6 7, your prediction did not go well. also it may worth checking number of rows in your submission.csv file.", "687295": "start with a public kernel and only modify the predictions based on your algorithm." }, "source": "meta" }