{
  "id": 482527,
  "title": "Can I assume eeg_id is unique in actual test.csv?",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/482527",
  "author_name": "",
  "post_date": "2024-03-08T04:05:32.435991900Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>My submission fails though Notebook is successful. My data prep code and final prediction file build runs based on the assumption that eeg_id is unique in test.csv. I am not sure if the submission file does not pass validation due to the consequences of this assumption. As the predictions in submission file are required by eeg_id, I assume this should be unique in actual test.csv file. Can anyone confirm?</p>",
  "messages": [
    {
      "id": "2686765",
      "postDate": "03/08/2024 04:05:32",
      "content": "<p>My submission fails though Notebook is successful. My data prep code and final prediction file build runs based on the assumption that eeg_id is unique in test.csv. I am not sure if the submission file does not pass validation due to the consequences of this assumption. As the predictions in submission file are required by eeg_id, I assume this should be unique in actual test.csv file. Can anyone confirm?</p>",
      "rawMarkdown": "My submission fails though Notebook is successful. My data prep code and final prediction file build runs based on the assumption that eeg_id is unique in test.csv. I am not sure if the submission file does not pass validation due to the consequences of this assumption. As the predictions in submission file are required by eeg_id, I assume this should be unique in actual test.csv file. Can anyone confirm?",
      "votes": null
    },
    {
      "id": "2686768",
      "postDate": "03/08/2024 04:09:34",
      "content": "<p>Yes i believe <code>eeg_id</code> is unique in test set. This is the column that Kaggle LB uses to score our submissions (therefore it needs to be unique).</p>",
      "rawMarkdown": "Yes i believe `eeg_id` is unique in test set. This is the column that Kaggle LB uses to score our submissions (therefore it needs to be unique).",
      "votes": null
    },
    {
      "id": "2687691",
      "postDate": "03/08/2024 17:38:16",
      "content": "<p>Chris is correct. Whatever ID column gets used in the sample submission is unique in most competitions, with very rare exceptions.</p>",
      "rawMarkdown": "Chris is correct. Whatever ID column gets used in the sample submission is unique in most competitions, with very rare exceptions.",
      "votes": null
    },
    {
      "id": "2688073",
      "postDate": "03/09/2024 01:28:14",
      "content": "<p>Maybe your failure is due to softmax problem, using torch.softmax could solve it.</p>",
      "rawMarkdown": "Maybe your failure is due to softmax problem, using torch.softmax could solve it.",
      "votes": null
    },
    {
      "id": "2689678",
      "postDate": "03/10/2024 04:22:45",
      "content": "<p>I know that patient_id is not unique if that helps</p>",
      "rawMarkdown": "I know that patient_id is not unique if that helps",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2686768,
      "author_name": "cdeotte",
      "author_url": "",
      "post_date": "03/08/2024 04:09:34",
      "content": "<p>Yes i believe <code>eeg_id</code> is unique in test set. This is the column that Kaggle LB uses to score our submissions (therefore it needs to be unique).</p>",
      "votes": null,
      "replies": [
        {
          "id": 2687691,
          "author_name": "sohier",
          "author_url": "",
          "post_date": "03/08/2024 17:38:16",
          "content": "<p>Chris is correct. Whatever ID column gets used in the sample submission is unique in most competitions, with very rare exceptions.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2688073,
      "author_name": "goldenlock",
      "author_url": "",
      "post_date": "03/09/2024 01:28:14",
      "content": "<p>Maybe your failure is due to softmax problem, using torch.softmax could solve it.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2689678,
      "author_name": "chrisrichardmiles",
      "author_url": "",
      "post_date": "03/10/2024 04:22:45",
      "content": "<p>I know that patient_id is not unique if that helps</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2686765": "My submission fails though Notebook is successful. My data prep code and final prediction file build runs based on the assumption that eeg_id is unique in test.csv. I am not sure if the submission file does not pass validation due to the consequences of this assumption. As the predictions in submission file are required by eeg_id, I assume this should be unique in actual test.csv file. Can anyone confirm?",
    "2686768": "Yes i believe `eeg_id` is unique in test set. This is the column that Kaggle LB uses to score our submissions (therefore it needs to be unique).",
    "2687691": "Chris is correct. Whatever ID column gets used in the sample submission is unique in most competitions, with very rare exceptions.",
    "2688073": "Maybe your failure is due to softmax problem, using torch.softmax could solve it.",
    "2689678": "I know that patient_id is not unique if that helps"
  },
  "source": "meta"
}