{
  "id": 99768,
  "title": "Getting Error when submitting as csv files",
  "url": "/competitions/aptos2019-blindness-detection/discussion/99768",
  "author_name": "Ali Haidar",
  "post_date": "2019-07-14T04:08:58.138000",
  "votes": 0,
  "comment_count": 7,
  "views": 0,
  "content": "<p>I ran my model offline then uploaded the results as submission.csv</p>\n\n<p><code>import numpy as np \nimport pandas as pd \nimport os\nsub=pd.read_csv('../input/submission.csv')\nsub.to_csv('submission.csv',index=None)</code></p>\n\n<p>The csv file is generated correctly. However, when submitting the file to competition, i am getting  Error keyword instead of score. Just wondering if someone else faced the same issue and solved it.</p>",
  "messages": [
    {
      "id": 574547,
      "postDate": "2019-07-14T04:08:58.140Z",
      "content": "<p>I ran my model offline then uploaded the results as submission.csv</p>\n\n<p><code>import numpy as np \nimport pandas as pd \nimport os\nsub=pd.read_csv('../input/submission.csv')\nsub.to_csv('submission.csv',index=None)</code></p>\n\n<p>The csv file is generated correctly. However, when submitting the file to competition, i am getting  Error keyword instead of score. Just wondering if someone else faced the same issue and solved it.</p>",
      "rawMarkdown": "I ran my model offline then uploaded the results as submission.csv\n\n`import numpy as np \nimport pandas as pd \nimport os\nsub=pd.read_csv('../input/submission.csv')\nsub.to_csv('submission.csv',index=None)`\n\nThe csv file is generated correctly. However, when submitting the file to competition, i am getting  Error keyword instead of score. Just wondering if someone else faced the same issue and solved it."
    },
    {
      "id": 575067,
      "postDate": "2019-07-15T02:18:37.427Z",
      "content": "<p>You mean that I can't simply upload through csv ? all the models should be built in kernals?</p>",
      "rawMarkdown": "You mean that I can't simply upload through csv ? all the models should be built in kernals?",
      "replies": [
        {
          "id": 575283,
          "postDate": "2019-07-15T08:37:08.310Z",
          "content": "<p>Model can be trained outside. But for prediction model should be present in Kernel  as there is hidden test set.</p>",
          "rawMarkdown": "Model can be trained outside. But for prediction model should be present in Kernel  as there is hidden test set.",
          "votes": 1
        },
        {
          "id": 575314,
          "postDate": "2019-07-15T09:08:38.320Z",
          "content": "<p>So, I should upload the model to the kernal correct ?</p>",
          "rawMarkdown": "So, I should upload the model to the kernal correct ?"
        },
        {
          "id": 575336,
          "postDate": "2019-07-15T09:52:49.283Z",
          "content": "<p>Yes. you should upload your model and write inference scripts to use that model on the test images and generate the submission.csv file, when you submit your kernel after commiting, Kaggle reruns your kernel on a hidden test set to compute the LB score.</p>",
          "rawMarkdown": "Yes. you should upload your model and write inference scripts to use that model on the test images and generate the submission.csv file, when you submit your kernel after commiting, Kaggle reruns your kernel on a hidden test set to compute the LB score.",
          "votes": 1
        },
        {
          "id": 575479,
          "postDate": "2019-07-15T13:18:32.250Z",
          "content": "<p>thanks very much</p>",
          "rawMarkdown": "thanks very much"
        }
      ]
    },
    {
      "id": 574664,
      "postDate": "2019-07-14T09:18:49.667Z",
      "content": "<p>You have to run model in kernel only. On submission, model will run on private test set also.</p>",
      "rawMarkdown": "You have to run model in kernel only. On submission, model will run on private test set also."
    },
    {
      "id": 575087,
      "postDate": "2019-07-15T03:21:50.033Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 575067,
      "author_name": "Ali Haidar",
      "author_url": "",
      "post_date": "2019-07-15T02:18:37.427000",
      "content": "<p>You mean that I can't simply upload through csv ? all the models should be built in kernals?</p>",
      "votes": 0,
      "replies": [
        {
          "id": 575283,
          "author_name": "Murlikrishnan Tripathi",
          "author_url": "",
          "post_date": "2019-07-15T08:37:08.310000",
          "content": "<p>Model can be trained outside. But for prediction model should be present in Kernel  as there is hidden test set.</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 575314,
          "author_name": "Ali Haidar",
          "author_url": "",
          "post_date": "2019-07-15T09:08:38.320000",
          "content": "<p>So, I should upload the model to the kernal correct ?</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 575336,
          "author_name": "Rishabh Agrahari",
          "author_url": "",
          "post_date": "2019-07-15T09:52:49.283000",
          "content": "<p>Yes. you should upload your model and write inference scripts to use that model on the test images and generate the submission.csv file, when you submit your kernel after commiting, Kaggle reruns your kernel on a hidden test set to compute the LB score.</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 575479,
          "author_name": "Ali Haidar",
          "author_url": "",
          "post_date": "2019-07-15T13:18:32.250000",
          "content": "<p>thanks very much</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 574664,
      "author_name": "Murlikrishnan Tripathi",
      "author_url": "",
      "post_date": "2019-07-14T09:18:49.667000",
      "content": "<p>You have to run model in kernel only. On submission, model will run on private test set also.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 575087,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-07-15T03:21:50.033000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "574547": "I ran my model offline then uploaded the results as submission.csv\n\n`import numpy as np \nimport pandas as pd \nimport os\nsub=pd.read_csv('../input/submission.csv')\nsub.to_csv('submission.csv',index=None)`\n\nThe csv file is generated correctly. However, when submitting the file to competition, i am getting  Error keyword instead of score. Just wondering if someone else faced the same issue and solved it.",
    "575067": "You mean that I can't simply upload through csv ? all the models should be built in kernals?",
    "574664": "You have to run model in kernel only. On submission, model will run on private test set also.",
    "575087": ""
  }
}