{
  "id": 471869,
  "title": "patients in test data",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/471869",
  "author_name": "",
  "post_date": "2024-01-29T22:13:31.510495300Z",
  "votes": 6,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Apologies if this has already been answered. Will the final evaluation data be for patients not present in the training data? </p>",
  "messages": [
    {
      "id": "2626312",
      "postDate": "01/29/2024 22:13:31",
      "content": "<p>Apologies if this has already been answered. Will the final evaluation data be for patients not present in the training data? </p>",
      "rawMarkdown": "Apologies if this has already been answered. Will the final evaluation data be for patients not present in the training data?",
      "votes": null
    },
    {
      "id": "2626317",
      "postDate": "01/29/2024 22:18:58",
      "content": "<p>Nobody has answered this question yet. We should probe the test data with an <code>if-then</code> test. If test patients in train, then submit preds = 1/6, else submit preds = train means. Then we will know.</p>",
      "rawMarkdown": "Nobody has answered this question yet. We should probe the test data with an `if-then` test. If test patients in train, then submit preds = 1/6, else submit preds = train means. Then we will know.",
      "votes": null
    },
    {
      "id": "2626333",
      "postDate": "01/29/2024 22:36:43",
      "content": "<p>Thanks. Reason I ask is that I am reading \"Regularization of Deep Neural Networks for EEG Seizure<br>\nDetection to Mitigate Overfitting\". <br>\nMohammed Saqib1,Yuanda Zhu2, May Dongmei Wang1, Brett Beaulieu-Jones3 It seems that detecting seizures for a patient in the training data is quite accurate, but a new patient is much more difficult. </p>",
      "rawMarkdown": "Thanks. Reason I ask is that I am reading \"Regularization of Deep Neural Networks for EEG Seizure\nDetection to Mitigate Overfitting\". \nMohammed Saqib1,Yuanda Zhu2, May Dongmei Wang1, Brett Beaulieu-Jones3 It seems that detecting seizures for a patient in the training data is quite accurate, but a new patient is much more difficult.",
      "votes": null
    },
    {
      "id": "2626337",
      "postDate": "01/29/2024 22:41:33",
      "content": "<p><a href=\"https://pubmed.ncbi.nlm.nih.gov/33073266/\" target=\"_blank\">https://pubmed.ncbi.nlm.nih.gov/33073266/</a></p>",
      "rawMarkdown": "https://pubmed.ncbi.nlm.nih.gov/33073266/",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2626317,
      "author_name": "cdeotte",
      "author_url": "",
      "post_date": "01/29/2024 22:18:58",
      "content": "<p>Nobody has answered this question yet. We should probe the test data with an <code>if-then</code> test. If test patients in train, then submit preds = 1/6, else submit preds = train means. Then we will know.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2626333,
          "author_name": "ajenningsfrankston",
          "author_url": "",
          "post_date": "01/29/2024 22:36:43",
          "content": "<p>Thanks. Reason I ask is that I am reading \"Regularization of Deep Neural Networks for EEG Seizure<br>\nDetection to Mitigate Overfitting\". <br>\nMohammed Saqib1,Yuanda Zhu2, May Dongmei Wang1, Brett Beaulieu-Jones3 It seems that detecting seizures for a patient in the training data is quite accurate, but a new patient is much more difficult. </p>",
          "votes": null,
          "replies": [
            {
              "id": 2626337,
              "author_name": "ajenningsfrankston",
              "author_url": "",
              "post_date": "01/29/2024 22:41:33",
              "content": "<p><a href=\"https://pubmed.ncbi.nlm.nih.gov/33073266/\" target=\"_blank\">https://pubmed.ncbi.nlm.nih.gov/33073266/</a></p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2626312": "Apologies if this has already been answered. Will the final evaluation data be for patients not present in the training data?",
    "2626317": "Nobody has answered this question yet. We should probe the test data with an `if-then` test. If test patients in train, then submit preds = 1/6, else submit preds = train means. Then we will know.",
    "2626333": "Thanks. Reason I ask is that I am reading \"Regularization of Deep Neural Networks for EEG Seizure\nDetection to Mitigate Overfitting\". \nMohammed Saqib1,Yuanda Zhu2, May Dongmei Wang1, Brett Beaulieu-Jones3 It seems that detecting seizures for a patient in the training data is quite accurate, but a new patient is much more difficult.",
    "2626337": "https://pubmed.ncbi.nlm.nih.gov/33073266/"
  },
  "source": "meta"
}