{
  "id": 106084,
  "title": "blending too much means overfit????",
  "url": "/competitions/champs-scalar-coupling/discussion/106084",
  "author_name": "",
  "post_date": "2019-08-28T05:00:36.274585800Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>i'm not sure it's just fit the LB ,because i use weight blend by its LB socre and feature used. \nmodel's LB and CV is much lower.All below -1.67 .</p>",
  "messages": [
    {
      "id": "609771",
      "postDate": "08/28/2019 05:00:36",
      "content": "<p>i'm not sure it's just fit the LB ,because i use weight blend by its LB socre and feature used. \nmodel's LB and CV is much lower.All below -1.67 .</p>",
      "rawMarkdown": "i'm not sure it's just fit the LB ,because i use weight blend by its LB socre and feature used. \nmodel's LB and CV is much lower.All below -1.67 .",
      "votes": null
    },
    {
      "id": "609775",
      "postDate": "08/28/2019 05:04:48",
      "content": "<blockquote>\n  <p>weight blend by its LB socre</p>\n</blockquote>\n\n<p>Each time you use LB score to guide model selection then you fit to public LB, and risk overfitting.</p>",
      "rawMarkdown": "&gt; weight blend by its LB socre\n\nEach time you use LB score to guide model selection then you fit to public LB, and risk overfitting.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 609775,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "08/28/2019 05:04:48",
      "content": "<blockquote>\n  <p>weight blend by its LB socre</p>\n</blockquote>\n\n<p>Each time you use LB score to guide model selection then you fit to public LB, and risk overfitting.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "609771": "i'm not sure it's just fit the LB ,because i use weight blend by its LB socre and feature used. \nmodel's LB and CV is much lower.All below -1.67 .",
    "609775": "&gt; weight blend by its LB socre\n\nEach time you use LB score to guide model selection then you fit to public LB, and risk overfitting."
  },
  "source": "meta"
}