{
  "id": 9523,
  "title": "chain 284 has only one week of data",
  "url": "/competitions/acquire-valued-shoppers-challenge/discussion/9523",
  "author_name": "",
  "post_date": "2014-06-23T17:39:12.887Z",
  "votes": null,
  "comment_count": 3,
  "views": 1091,
  "content": "<p>Info provided says we are given a minimum of a year of shopping history prior to the incentive. But all the transaction data for chain 284 occurred between 2013-06-17 and 2013-06-23 and there are 14 customers (all from the test file) nested in chain 284. I suspect this was a typo and it should be 285, but that id was assigned within chain after the typo was made.</p>\n<p>If this anomaly occurred in the training file I would probably just throw this data out, but the customers are in the test set so it appears that we we must supply predictions for them without having the promised year of data. Hmm...</p>",
  "messages": [
    {
      "id": "49428",
      "postDate": "06/23/2014 17:39:12",
      "content": "<p>Info provided says we are given a minimum of a year of shopping history prior to the incentive. But all the transaction data for chain 284 occurred between 2013-06-17 and 2013-06-23 and there are 14 customers (all from the test file) nested in chain 284. I suspect this was a typo and it should be 285, but that id was assigned within chain after the typo was made.</p>\n<p>If this anomaly occurred in the training file I would probably just throw this data out, but the customers are in the test set so it appears that we we must supply predictions for them without having the promised year of data. Hmm...</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "49446",
      "postDate": "06/24/2014 01:52:55",
      "content": "<p>I have a similar question about how to predict those customers that have nearly &quot;zero&quot; history.&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "49450",
      "postDate": "06/24/2014 04:22:25",
      "content": "<p>A year of shopping data for an individual may have only one transaction in that year.</p>\n\n<p>So a transaction January 1, 2013, and they give you all the data for 2013, would only have that one transaction.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "49451",
      "postDate": "06/24/2014 04:50:06",
      "content": "<p>Yes, I realize that, and I believe that is the explanation for some customers with sparse data. But this is a *chain* that only has one week of data. Since customers are nested within chain, all customers nested in this chain (about 14) also only have one week of data.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 49446,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "06/24/2014 01:52:55",
      "content": "<p>I have a similar question about how to predict those customers that have nearly &quot;zero&quot; history.&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 49450,
      "author_name": "letterrip",
      "author_url": "",
      "post_date": "06/24/2014 04:22:25",
      "content": "<p>A year of shopping data for an individual may have only one transaction in that year.</p>\n\n<p>So a transaction January 1, 2013, and they give you all the data for 2013, would only have that one transaction.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 49451,
      "author_name": "swhardy",
      "author_url": "",
      "post_date": "06/24/2014 04:50:06",
      "content": "<p>Yes, I realize that, and I believe that is the explanation for some customers with sparse data. But this is a *chain* that only has one week of data. Since customers are nested within chain, all customers nested in this chain (about 14) also only have one week of data.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "49428": "",
    "49446": "",
    "49450": "",
    "49451": ""
  },
  "source": "meta"
}