{
  "id": 404989,
  "title": "Is tasks.csv not included for defog tests? ",
  "url": "/competitions/tlvmc-parkinsons-freezing-gait-prediction/discussion/404989",
  "author_name": "",
  "post_date": "2023-04-25T18:58:31.874309800Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I found other metadata is updated for the test files but not the tasks.csv. I just used this code for submission and the code will fail. </p>\n<pre><code>test_files = glob.glob()\ntasks = pd.read_csv()\n\n file  test_files:\n     file.find()&gt;:\n        file_Id = file.split()[-].split()[]        \n         file_Id  tasks[].unique().tolist() \n</code></pre>\n<p>I assume the tasks table is important metadata for defog files. Has anyone found this problem? Thanks.</p>",
  "messages": [
    {
      "id": "2235141",
      "postDate": "04/25/2023 18:58:31",
      "content": "<p>I found other metadata is updated for the test files but not the tasks.csv. I just used this code for submission and the code will fail. </p>\n<pre><code>test_files = glob.glob()\ntasks = pd.read_csv()\n\n file  test_files:\n     file.find()&gt;:\n        file_Id = file.split()[-].split()[]        \n         file_Id  tasks[].unique().tolist() \n</code></pre>\n<p>I assume the tasks table is important metadata for defog files. Has anyone found this problem? Thanks.</p>",
      "rawMarkdown": "I found other metadata is updated for the test files but not the tasks.csv. I just used this code for submission and the code will fail. \n```python\ntest_files = glob.glob(f'{DIR}/test/**/**')\ntasks = pd.read_csv(f\"{DIR}/tasks.csv\")\n\nfor file in test_files:\n    if file.find('defog')>0:\n        file_Id = file.split('/')[-1].split('.')[0]        \n        assert file_Id in tasks['Id'].unique().tolist() \n``` I assume the tasks table is important metadata for defog files. Has anyone found this problem? Thanks.",
      "votes": null
    },
    {
      "id": "2247179",
      "postDate": "05/05/2023 19:22:12",
      "content": "<p><a href=\"https://www.kaggle.com/ryanholbrook\" target=\"_blank\">@ryanholbrook</a> Can you confirm that the tasks are not updated for the tests submission? Thanks. </p>",
      "rawMarkdown": "ryanholbrook Can you confirm that the tasks are not updated for the tests submission? Thanks.",
      "votes": null
    },
    {
      "id": "2247209",
      "postDate": "05/05/2023 19:55:22",
      "content": "<p>That's correct. We excluded the tasks in the test set to prevent data leakage. The metric is aware of the task locations, however, and will ignore any predictions outside of them.</p>",
      "rawMarkdown": "That's correct. We excluded the tasks in the test set to prevent data leakage. The metric is aware of the task locations, however, and will ignore any predictions outside of them.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2247179,
      "author_name": "ningjiaca",
      "author_url": "",
      "post_date": "05/05/2023 19:22:12",
      "content": "<p><a href=\"https://www.kaggle.com/ryanholbrook\" target=\"_blank\">@ryanholbrook</a> Can you confirm that the tasks are not updated for the tests submission? Thanks. </p>",
      "votes": null,
      "replies": [
        {
          "id": 2247209,
          "author_name": "ryanholbrook",
          "author_url": "",
          "post_date": "05/05/2023 19:55:22",
          "content": "<p>That's correct. We excluded the tasks in the test set to prevent data leakage. The metric is aware of the task locations, however, and will ignore any predictions outside of them.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2235141": "I found other metadata is updated for the test files but not the tasks.csv. I just used this code for submission and the code will fail. \n```python\ntest_files = glob.glob(f'{DIR}/test/**/**')\ntasks = pd.read_csv(f\"{DIR}/tasks.csv\")\n\nfor file in test_files:\n    if file.find('defog')>0:\n        file_Id = file.split('/')[-1].split('.')[0]        \n        assert file_Id in tasks['Id'].unique().tolist() \n``` I assume the tasks table is important metadata for defog files. Has anyone found this problem? Thanks.",
    "2247179": "ryanholbrook Can you confirm that the tasks are not updated for the tests submission? Thanks.",
    "2247209": "That's correct. We excluded the tasks in the test set to prevent data leakage. The metric is aware of the task locations, however, and will ignore any predictions outside of them."
  },
  "source": "meta"
}