{
  "id": 287524,
  "title": "Submission Scoring Error",
  "url": "/competitions/g-research-crypto-forecasting/discussion/287524",
  "author_name": "",
  "post_date": "2021-11-14T13:05:14.579379700Z",
  "votes": 11,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi all,</p>\n<p>I'm trying to submit but get the above error. I can submit fine when I set all \"Target\" equal to a real number but when I set sample_prediction_df['Target'] = PredArray, where PredArray is a numpy array of the predicted targets I get a \"Submission Scoring Error\". The csv files are identical, I've checked the type on the dataframes and numbers being submitted and can't seem to find where the error is coming from. </p>\n<p>I was wondering if anybody knows what could be going wrong here?</p>",
  "messages": [
    {
      "id": "1582019",
      "postDate": "11/14/2021 13:05:14",
      "content": "<p>Hi all,</p>\n<p>I'm trying to submit but get the above error. I can submit fine when I set all \"Target\" equal to a real number but when I set sample_prediction_df['Target'] = PredArray, where PredArray is a numpy array of the predicted targets I get a \"Submission Scoring Error\". The csv files are identical, I've checked the type on the dataframes and numbers being submitted and can't seem to find where the error is coming from. </p>\n<p>I was wondering if anybody knows what could be going wrong here?</p>",
      "rawMarkdown": "Hi all,\n\nI'm trying to submit but get the above error. I can submit fine when I set all \"Target\" equal to a real number but when I set sample_prediction_df['Target'] = PredArray, where PredArray is a numpy array of the predicted targets I get a \"Submission Scoring Error\". The csv files are identical, I've checked the type on the dataframes and numbers being submitted and can't seem to find where the error is coming from. \n\nI was wondering if anybody knows what could be going wrong here?",
      "votes": null
    },
    {
      "id": "1582440",
      "postDate": "11/14/2021 22:28:10",
      "content": "<p>It happens because When the API run in private data sometimes not all of Asset exist. sometimes there is no data for some crypto</p>",
      "rawMarkdown": "It happens because When the API run in private data sometimes not all of Asset exist. sometimes there is no data for some crypto",
      "votes": null
    },
    {
      "id": "1582622",
      "postDate": "11/15/2021 05:10:35",
      "content": "<p>Lets explain how kaggle scoring works there is a public lb and also a private lb public lb is your current score and the private lb is identical but on a hidden dataset the score there is revealed in the end and your final rank is what is shown there . When you submit it runs your code on a hidden private test and produces hidden submission on that let's say public lb has j samples and private lb has i samples when you run your code it expects to produce i length submission but as you are not doing it on the private set and the public set only it produces j samples thus it fails the test csv is updated with a private set only when you submit and your code is rerun with i samples and expects your model to predict on that you will never be able to find the hidden test csv with i samples but will be able to find the one with j samples </p>",
      "rawMarkdown": "Lets explain how kaggle scoring works there is a public lb and also a private lb public lb is your current score and the private lb is identical but on a hidden dataset the score there is revealed in the end and your final rank is what is shown there . When you submit it runs your code on a hidden private test and produces hidden submission on that let's say public lb has j samples and private lb has i samples when you run your code it expects to produce i length submission but as you are not doing it on the private set and the public set only it produces j samples thus it fails the test csv is updated with a private set only when you submit and your code is rerun with i samples and expects your model to predict on that you will never be able to find the hidden test csv with i samples but will be able to find the one with j samples",
      "votes": null
    },
    {
      "id": "1626566",
      "postDate": "12/23/2021 01:50:29",
      "content": "<p>What would be a potential solution?</p>",
      "rawMarkdown": "What would be a potential solution?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1582440,
      "author_name": "ebudia",
      "author_url": "",
      "post_date": "11/14/2021 22:28:10",
      "content": "<p>It happens because When the API run in private data sometimes not all of Asset exist. sometimes there is no data for some crypto</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1582622,
      "author_name": "swaralipibose",
      "author_url": "",
      "post_date": "11/15/2021 05:10:35",
      "content": "<p>Lets explain how kaggle scoring works there is a public lb and also a private lb public lb is your current score and the private lb is identical but on a hidden dataset the score there is revealed in the end and your final rank is what is shown there . When you submit it runs your code on a hidden private test and produces hidden submission on that let's say public lb has j samples and private lb has i samples when you run your code it expects to produce i length submission but as you are not doing it on the private set and the public set only it produces j samples thus it fails the test csv is updated with a private set only when you submit and your code is rerun with i samples and expects your model to predict on that you will never be able to find the hidden test csv with i samples but will be able to find the one with j samples </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1626566,
      "author_name": "mdominguez2010",
      "author_url": "",
      "post_date": "12/23/2021 01:50:29",
      "content": "<p>What would be a potential solution?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1582019": "Hi all,\n\nI'm trying to submit but get the above error. I can submit fine when I set all \"Target\" equal to a real number but when I set sample_prediction_df['Target'] = PredArray, where PredArray is a numpy array of the predicted targets I get a \"Submission Scoring Error\". The csv files are identical, I've checked the type on the dataframes and numbers being submitted and can't seem to find where the error is coming from. \n\nI was wondering if anybody knows what could be going wrong here?",
    "1582440": "It happens because When the API run in private data sometimes not all of Asset exist. sometimes there is no data for some crypto",
    "1582622": "Lets explain how kaggle scoring works there is a public lb and also a private lb public lb is your current score and the private lb is identical but on a hidden dataset the score there is revealed in the end and your final rank is what is shown there . When you submit it runs your code on a hidden private test and produces hidden submission on that let's say public lb has j samples and private lb has i samples when you run your code it expects to produce i length submission but as you are not doing it on the private set and the public set only it produces j samples thus it fails the test csv is updated with a private set only when you submit and your code is rerun with i samples and expects your model to predict on that you will never be able to find the hidden test csv with i samples but will be able to find the one with j samples",
    "1626566": "What would be a potential solution?"
  },
  "source": "meta"
}