{
  "id": 53675,
  "title": "Validation Strategies",
  "url": "/competitions/data-science-bowl-2018/discussion/53675",
  "author_name": "",
  "post_date": "2018-04-03T12:11:25.892804900Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Wondering if anyone had any robust validation strategies or at least validation results consistent with LB. </p>",
  "messages": [
    {
      "id": "308360",
      "postDate": "04/03/2018 12:11:25",
      "content": "<p>Wondering if anyone had any robust validation strategies or at least validation results consistent with LB. </p>",
      "rawMarkdown": "Wondering if anyone had any robust validation strategies or at least validation results consistent with LB.",
      "votes": null
    },
    {
      "id": "308422",
      "postDate": "04/03/2018 13:46:48",
      "content": "<p>I saw Heng CherKeng validate on 43 of the gray images when training a gray image model, but when doing so the mean average precision varied from 0.63 to 0.69 making it useless to try and choose what iteration would yield the best model. </p>",
      "rawMarkdown": "I saw Heng CherKeng validate on 43 of the gray images when training a gray image model, but when doing so the mean average precision varied from 0.63 to 0.69 making it useless to try and choose what iteration would yield the best model.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 308422,
      "author_name": "livingprogram",
      "author_url": "",
      "post_date": "04/03/2018 13:46:48",
      "content": "<p>I saw Heng CherKeng validate on 43 of the gray images when training a gray image model, but when doing so the mean average precision varied from 0.63 to 0.69 making it useless to try and choose what iteration would yield the best model. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "308360": "Wondering if anyone had any robust validation strategies or at least validation results consistent with LB.",
    "308422": "I saw Heng CherKeng validate on 43 of the gray images when training a gray image model, but when doing so the mean average precision varied from 0.63 to 0.69 making it useless to try and choose what iteration would yield the best model."
  },
  "source": "meta"
}