{
  "id": 44482,
  "title": "Moving Average: How to deal with NA MA values",
  "url": "/competitions/favorita-grocery-sales-forecasting/discussion/44482",
  "author_name": "",
  "post_date": "2017-11-29T10:00:50.517857700Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I noticed that some items doesn’t have sales for certain days and that makes it hard to calculate an accurate MA. How did you guys get around this problem?</p>",
  "messages": [
    {
      "id": "249842",
      "postDate": "11/29/2017 10:00:50",
      "content": "<p>I noticed that some items doesn’t have sales for certain days and that makes it hard to calculate an accurate MA. How did you guys get around this problem?</p>",
      "rawMarkdown": "I noticed that some items doesn’t have sales for certain days and that makes it hard to calculate an accurate MA. How did you guys get around this problem?",
      "votes": null
    },
    {
      "id": "250027",
      "postDate": "11/29/2017 16:52:27",
      "content": "<p>Most of the kernels fill in the missing dates, and assume unit_sales and onpromotion are 0. <br>\nSo if your train set is 2017-01-01 --&gt; 2017-08-15, then each item/store combination will have an entry for every date.  </p>",
      "rawMarkdown": "Most of the kernels fill in the missing dates, and assume unit_sales and onpromotion are 0.  \nSo if your train set is 2017-01-01 --&gt; 2017-08-15, then each item/store combination will have an entry for every date.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 250027,
      "author_name": "npa02012",
      "author_url": "",
      "post_date": "11/29/2017 16:52:27",
      "content": "<p>Most of the kernels fill in the missing dates, and assume unit_sales and onpromotion are 0. <br>\nSo if your train set is 2017-01-01 --&gt; 2017-08-15, then each item/store combination will have an entry for every date.  </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "249842": "I noticed that some items doesn’t have sales for certain days and that makes it hard to calculate an accurate MA. How did you guys get around this problem?",
    "250027": "Most of the kernels fill in the missing dates, and assume unit_sales and onpromotion are 0.  \nSo if your train set is 2017-01-01 --&gt; 2017-08-15, then each item/store combination will have an entry for every date."
  },
  "source": "meta"
}