{
  "id": 12413,
  "title": "how to set up this data set ?",
  "url": "/competitions/avazu-ctr-prediction/discussion/12413",
  "author_name": "kayama",
  "post_date": "2015-02-05T00:42:02.337000",
  "votes": 0,
  "comment_count": 0,
  "views": 570,
  "content": "<p>Hi All,</p>\n<p>I am new to ML &amp; does not make it to submit the first trial in time...</p>\n<p>My question is how do I train this data set as it seems like to me the behavior of the user should based on the device ID</p>\n<p>So if it is true we would like to train the data set e.g. Device ID=12345</p>\n<p>based on the behavior pattern of feature1..featureM</p>\n<p>but now this is a big file with million of DeviceID. So how I can tell my algorithm to train them as such ?</p>\n<p>Or I need to pre-process the big file and make it into N chunks and each chunks corresponds to that particular user...and train for that N chunks and predict for each user ?&nbsp;If so, it will end up with lots of files &nbsp;to train..?</p>\n<p>It seems some what in-efficient..</p>\n<p>What is the better practice for this problem?</p>",
  "messages": [
    {
      "id": 63621,
      "postDate": "2015-02-05T00:42:02.337Z",
      "content": "<p>Hi All,</p>\n<p>I am new to ML &amp; does not make it to submit the first trial in time...</p>\n<p>My question is how do I train this data set as it seems like to me the behavior of the user should based on the device ID</p>\n<p>So if it is true we would like to train the data set e.g. Device ID=12345</p>\n<p>based on the behavior pattern of feature1..featureM</p>\n<p>but now this is a big file with million of DeviceID. So how I can tell my algorithm to train them as such ?</p>\n<p>Or I need to pre-process the big file and make it into N chunks and each chunks corresponds to that particular user...and train for that N chunks and predict for each user ?&nbsp;If so, it will end up with lots of files &nbsp;to train..?</p>\n<p>It seems some what in-efficient..</p>\n<p>What is the better practice for this problem?</p>"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "63621": ""
  }
}