{
  "id": 56101,
  "title": "Chunk Learning vs One Time Training",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/56101",
  "author_name": "Jason Chan",
  "post_date": "2018-05-06T03:36:44.667000",
  "votes": 0,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n\n<p>Just a curious question I had in mind. Has there been someone out there who can testify which gives better LB scores ? is it training by chunk learning on entire dataset or one time training on entire dataset ? Assuming training using the same model and parameters.</p>\n\n<p>Regards,\nJason</p>",
  "messages": [
    {
      "id": 324042,
      "postDate": "2018-05-07T02:33:28.133Z",
      "content": "<p>I have been trying Deep Learning methods for this competition in order to get hands on experience on Deep Learning.\nAs you can see my scores are not good.\nFor me chunk learning gave me better score than entire dataset at once. I am also looking a reason for that behavior </p>",
      "rawMarkdown": "I have been trying Deep Learning methods for this competition in order to get hands on experience on Deep Learning.\nAs you can see my scores are not good.\nFor me chunk learning gave me better score than entire dataset at once. I am also looking a reason for that behavior "
    },
    {
      "id": 323741,
      "postDate": "2018-05-06T03:36:44.667Z",
      "content": "<p>Hi everyone,</p>\n\n<p>Just a curious question I had in mind. Has there been someone out there who can testify which gives better LB scores ? is it training by chunk learning on entire dataset or one time training on entire dataset ? Assuming training using the same model and parameters.</p>\n\n<p>Regards,\nJason</p>",
      "rawMarkdown": "Hi everyone,\n\nJust a curious question I had in mind. Has there been someone out there who can testify which gives better LB scores ? is it training by chunk learning on entire dataset or one time training on entire dataset ? Assuming training using the same model and parameters.\n\nRegards,\nJason"
    },
    {
      "id": 323768,
      "postDate": "2018-05-06T06:07:56.827Z",
      "rawMarkdown": "",
      "votes": 1,
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 324042,
      "author_name": "isaranja",
      "author_url": "",
      "post_date": "2018-05-07T02:33:28.133000",
      "content": "<p>I have been trying Deep Learning methods for this competition in order to get hands on experience on Deep Learning.\nAs you can see my scores are not good.\nFor me chunk learning gave me better score than entire dataset at once. I am also looking a reason for that behavior </p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 323768,
      "author_name": "",
      "author_url": "",
      "post_date": "2018-05-06T06:07:56.827000",
      "content": "",
      "votes": 1,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "324042": "I have been trying Deep Learning methods for this competition in order to get hands on experience on Deep Learning.\nAs you can see my scores are not good.\nFor me chunk learning gave me better score than entire dataset at once. I am also looking a reason for that behavior ",
    "323741": "Hi everyone,\n\nJust a curious question I had in mind. Has there been someone out there who can testify which gives better LB scores ? is it training by chunk learning on entire dataset or one time training on entire dataset ? Assuming training using the same model and parameters.\n\nRegards,\nJason",
    "323768": ""
  }
}