{
  "id": 482835,
  "title": "Negative samples in training set",
  "url": "/competitions/bugnist2024fgvc/discussion/482835",
  "author_name": "",
  "post_date": "2024-03-09T16:53:08.804257Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>In doing some EDA, I found several training volumes which appear to contain no bug. Seems like they only contain a cap or base of a specimen tube? Here is one example (maddi_1_090):<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7448560%2Fc50c21bf7aac53ce3f3ce4e07c5044c1%2FScreenshot%20from%202024-03-09%2008-48-45.png?generation=1710002947410171&amp;alt=media\"></p>\n<p>The negative samples I've found are:<br>\nmaddi_1_090<br>\nbank_2_026<br>\nboffel_2_159<br>\nbank_6_116<br>\nboffel_4_095<br>\nguld_5_114</p>\n<p>Very small % of training data, so likely not an issue. Just thought I'd point it out.</p>",
  "messages": [
    {
      "id": "2689068",
      "postDate": "03/09/2024 16:53:08",
      "content": "<p>In doing some EDA, I found several training volumes which appear to contain no bug. Seems like they only contain a cap or base of a specimen tube? Here is one example (maddi_1_090):<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7448560%2Fc50c21bf7aac53ce3f3ce4e07c5044c1%2FScreenshot%20from%202024-03-09%2008-48-45.png?generation=1710002947410171&amp;alt=media\"></p>\n<p>The negative samples I've found are:<br>\nmaddi_1_090<br>\nbank_2_026<br>\nboffel_2_159<br>\nbank_6_116<br>\nboffel_4_095<br>\nguld_5_114</p>\n<p>Very small % of training data, so likely not an issue. Just thought I'd point it out.</p>",
      "rawMarkdown": "In doing some EDA, I found several training volumes which appear to contain no bug. Seems like they only contain a cap or base of a specimen tube? Here is one example (maddi_1_090):\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7448560%2Fc50c21bf7aac53ce3f3ce4e07c5044c1%2FScreenshot%20from%202024-03-09%2008-48-45.png?generation=1710002947410171&alt=media)\n\nThe negative samples I've found are:\nmaddi_1_090\nbank_2_026\nboffel_2_159\nbank_6_116\nboffel_4_095\nguld_5_114\n\nVery small % of training data, so likely not an issue. Just thought I'd point it out.",
      "votes": null
    },
    {
      "id": "2689519",
      "postDate": "03/09/2024 22:22:13",
      "content": "<p>After being made aware of errors in the training data, I have removed 31 files and updated the data.</p>",
      "rawMarkdown": "After being made aware of errors in the training data, I have removed 31 files and updated the data.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2689519,
      "author_name": "andersbjorholmdahl",
      "author_url": "",
      "post_date": "03/09/2024 22:22:13",
      "content": "<p>After being made aware of errors in the training data, I have removed 31 files and updated the data.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2689068": "In doing some EDA, I found several training volumes which appear to contain no bug. Seems like they only contain a cap or base of a specimen tube? Here is one example (maddi_1_090):\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7448560%2Fc50c21bf7aac53ce3f3ce4e07c5044c1%2FScreenshot%20from%202024-03-09%2008-48-45.png?generation=1710002947410171&alt=media)\n\nThe negative samples I've found are:\nmaddi_1_090\nbank_2_026\nboffel_2_159\nbank_6_116\nboffel_4_095\nguld_5_114\n\nVery small % of training data, so likely not an issue. Just thought I'd point it out.",
    "2689519": "After being made aware of errors in the training data, I have removed 31 files and updated the data."
  },
  "source": "meta"
}