{
  "id": 112776,
  "title": "use of validation set",
  "url": "/competitions/understanding_cloud_organization/discussion/112776",
  "author_name": "",
  "post_date": "2019-10-15T09:59:43.649489600Z",
  "votes": -1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi, I have a question about the use of validation set after training on train set. If I don't use cv strategy, after seeing model performance on validation set, should i put the validation set into train set for more train epochs or just leave it alone? And which stop strategy should I use for training on the whole train set (including validation set)? Thanks!</p>",
  "messages": [
    {
      "id": "649407",
      "postDate": "10/15/2019 09:59:43",
      "content": "<p>Hi, I have a question about the use of validation set after training on train set. If I don't use cv strategy, after seeing model performance on validation set, should i put the validation set into train set for more train epochs or just leave it alone? And which stop strategy should I use for training on the whole train set (including validation set)? Thanks!</p>",
      "rawMarkdown": "Hi, I have a question about the use of validation set after training on train set. If I don't use cv strategy, after seeing model performance on validation set, should i put the validation set into train set for more train epochs or just leave it alone? And which stop strategy should I use for training on the whole train set (including validation set)? Thanks!",
      "votes": null
    },
    {
      "id": "651728",
      "postDate": "10/17/2019 21:21:36",
      "content": "<p>You can do cross_validation and obtain number epochs before overfitting and used this number epochs for prevent overfitting.</p>",
      "rawMarkdown": "You can do cross_validation and obtain number epochs before overfitting and used this number epochs for prevent overfitting.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 651728,
      "author_name": "miklgr500",
      "author_url": "",
      "post_date": "10/17/2019 21:21:36",
      "content": "<p>You can do cross_validation and obtain number epochs before overfitting and used this number epochs for prevent overfitting.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "649407": "Hi, I have a question about the use of validation set after training on train set. If I don't use cv strategy, after seeing model performance on validation set, should i put the validation set into train set for more train epochs or just leave it alone? And which stop strategy should I use for training on the whole train set (including validation set)? Thanks!",
    "651728": "You can do cross_validation and obtain number epochs before overfitting and used this number epochs for prevent overfitting."
  },
  "source": "meta"
}