{
  "id": 116725,
  "title": "Only one year of training data in train.csv?",
  "url": "/competitions/ashrae-energy-prediction/discussion/116725",
  "author_name": "",
  "post_date": "2019-11-11T03:16:36.606705300Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>The data set includes three years of hourly meter readings in the data description, but why is there only one year of training data in the given training set (train.csv) ?</p>",
  "messages": [
    {
      "id": "670119",
      "postDate": "11/11/2019 03:16:36",
      "content": "<p>The data set includes three years of hourly meter readings in the data description, but why is there only one year of training data in the given training set (train.csv) ?</p>",
      "rawMarkdown": "The data set includes three years of hourly meter readings in the data description, but why is there only one year of training data in the given training set (train.csv) ?",
      "votes": null
    },
    {
      "id": "670153",
      "postDate": "11/11/2019 05:06:20",
      "content": "<p><strong>2016 = Training set, 2017/2018 = Test set.</strong> They make that clear from the start, and please see the existing threads where this was already discussed. (Beware of the obvious dangers with only having one year as training set, such as entraining to specific public holiday dates (for the year 2016) and date-and-site-specific weather values).</p>",
      "rawMarkdown": "**2016 = Training set, 2017/2018 = Test set.** They make that clear from the start, and please see the existing threads where this was already discussed. (Beware of the obvious dangers with only having one year as training set, such as entraining to specific public holiday dates (for the year 2016) and date-and-site-specific weather values).",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 670153,
      "author_name": "smcinerney",
      "author_url": "",
      "post_date": "11/11/2019 05:06:20",
      "content": "<p><strong>2016 = Training set, 2017/2018 = Test set.</strong> They make that clear from the start, and please see the existing threads where this was already discussed. (Beware of the obvious dangers with only having one year as training set, such as entraining to specific public holiday dates (for the year 2016) and date-and-site-specific weather values).</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "670119": "The data set includes three years of hourly meter readings in the data description, but why is there only one year of training data in the given training set (train.csv) ?",
    "670153": "**2016 = Training set, 2017/2018 = Test set.** They make that clear from the start, and please see the existing threads where this was already discussed. (Beware of the obvious dangers with only having one year as training set, such as entraining to specific public holiday dates (for the year 2016) and date-and-site-specific weather values)."
  },
  "source": "meta"
}