{
  "id": 80415,
  "title": "Do using data augmentation always improve the model performance?",
  "url": "/competitions/histopathologic-cancer-detection/discussion/80415",
  "author_name": "",
  "post_date": "2019-02-13T11:55:43.589018600Z",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "",
  "messages": [
    {
      "id": "470698",
      "postDate": "02/13/2019 11:55:43",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "470721",
      "postDate": "02/13/2019 13:11:37",
      "content": "<p>Hi!\nFor what may concern my personal experience....I don't think so!\nIt depends on several parameters such as the kind of transformation you will apply and moreover from the results you want to predict ...and also from the dimension of your data. I think that data augmentation is very useful when you have a limited source of training data.</p>",
      "rawMarkdown": "Hi!\nFor what may concern my personal experience....I don't think so!\nIt depends on several parameters such as the kind of transformation you will apply and moreover from the results you want to predict ...and also from the dimension of your data. I think that data augmentation is very useful when you have a limited source of training data.",
      "votes": null
    },
    {
      "id": "470824",
      "postDate": "02/13/2019 16:15:06",
      "content": "<p>Yeah, I think the same way as you think <a href=\"/sparty87\">@sparty87</a></p>",
      "rawMarkdown": "Yeah, I think the same way as you think @sparty87",
      "votes": null
    },
    {
      "id": "474316",
      "postDate": "02/19/2019 07:50:22",
      "content": "<p>Hi,</p>\n\n<p>I believe data augmentation is usually used to address overfitting problems, if your model is experiencing underfitting or other types of problems(data leakage, etc..), data augmentation is not going to help you.</p>",
      "rawMarkdown": "Hi,\n\nI believe data augmentation is usually used to address overfitting problems, if your model is experiencing underfitting or other types of problems(data leakage, etc..), data augmentation is not going to help you.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 470721,
      "author_name": "sparty87",
      "author_url": "",
      "post_date": "02/13/2019 13:11:37",
      "content": "<p>Hi!\nFor what may concern my personal experience....I don't think so!\nIt depends on several parameters such as the kind of transformation you will apply and moreover from the results you want to predict ...and also from the dimension of your data. I think that data augmentation is very useful when you have a limited source of training data.</p>",
      "votes": null,
      "replies": [
        {
          "id": 470824,
          "author_name": "sparshdutta",
          "author_url": "",
          "post_date": "02/13/2019 16:15:06",
          "content": "<p>Yeah, I think the same way as you think <a href=\"/sparty87\">@sparty87</a></p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 474316,
      "author_name": "marsyeti",
      "author_url": "",
      "post_date": "02/19/2019 07:50:22",
      "content": "<p>Hi,</p>\n\n<p>I believe data augmentation is usually used to address overfitting problems, if your model is experiencing underfitting or other types of problems(data leakage, etc..), data augmentation is not going to help you.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "470698": "",
    "470721": "Hi!\nFor what may concern my personal experience....I don't think so!\nIt depends on several parameters such as the kind of transformation you will apply and moreover from the results you want to predict ...and also from the dimension of your data. I think that data augmentation is very useful when you have a limited source of training data.",
    "470824": "Yeah, I think the same way as you think @sparty87",
    "474316": "Hi,\n\nI believe data augmentation is usually used to address overfitting problems, if your model is experiencing underfitting or other types of problems(data leakage, etc..), data augmentation is not going to help you."
  },
  "source": "meta"
}