{
  "id": 39937,
  "title": "Can we use the expiration date in the members.csv? Is it future information?",
  "url": "/competitions/kkbox-churn-prediction-challenge/discussion/39937",
  "author_name": "",
  "post_date": "2017-09-24T11:25:18.506558700Z",
  "votes": 3,
  "comment_count": 5,
  "views": 0,
  "content": "",
  "messages": [
    {
      "id": "223942",
      "postDate": "09/24/2017 11:25:18",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "224056",
      "postDate": "09/25/2017 00:10:15",
      "content": "<p>I guess this should be equal to the last expiration date updated in the transactions until end of Feb... in such case it won't be forward looking. </p>",
      "rawMarkdown": "I guess this should be equal to the last expiration date updated in the transactions until end of Feb... in such case it won't be forward looking.",
      "votes": null
    },
    {
      "id": "224111",
      "postDate": "09/25/2017 06:31:41",
      "content": "<p>In my view, I think it's future information, I don't use it </p>",
      "rawMarkdown": "In my view, I think it's future information, I don't use it",
      "votes": null
    },
    {
      "id": "228793",
      "postDate": "10/07/2017 21:17:46",
      "content": "<p>No answer?</p>",
      "rawMarkdown": "No answer?",
      "votes": null
    },
    {
      "id": "232227",
      "postDate": "10/17/2017 06:35:12",
      "content": "<p>I think it's future information because it improve my OFFLINE model better than ONLINE.</p>",
      "rawMarkdown": "I think it's future information because it improve my OFFLINE model better than ONLINE.",
      "votes": null
    },
    {
      "id": "237881",
      "postDate": "10/31/2017 07:40:39",
      "content": "<p>I think it is future data. For example, you can see the first item in submission msno <code>'ugx0CjOMzazClkFzU2xasmDZaoIqOUAZPsH1q0teWCg='</code>. In the transactions.csv, the last transaction is on 20170228 with <code>membership_expire_date = 20170330</code> and <code>payment_plan_days = 30</code>. However, in <code>members.csv</code>, the <code>expiration_date</code> is <code>20170822</code>. From the transaction, you cannot know if the user has subscribe the service till 201708. And, if you know the user has expiration date on 201708, there is a very high chance that the user is not churn.</p>",
      "rawMarkdown": "I think it is future data. For example, you can see the first item in submission msno `'ugx0CjOMzazClkFzU2xasmDZaoIqOUAZPsH1q0teWCg='`. In the transactions.csv, the last transaction is on 20170228 with `membership_expire_date = 20170330` and `payment_plan_days = 30`. However, in `members.csv`, the `expiration_date` is `20170822`. From the transaction, you cannot know if the user has subscribe the service till 201708. And, if you know the user has expiration date on 201708, there is a very high chance that the user is not churn.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 224056,
      "author_name": "nicolasso",
      "author_url": "",
      "post_date": "09/25/2017 00:10:15",
      "content": "<p>I guess this should be equal to the last expiration date updated in the transactions until end of Feb... in such case it won't be forward looking. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 224111,
      "author_name": "franciszmy",
      "author_url": "",
      "post_date": "09/25/2017 06:31:41",
      "content": "<p>In my view, I think it's future information, I don't use it </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 228793,
      "author_name": "",
      "author_url": "",
      "post_date": "10/07/2017 21:17:46",
      "content": "<p>No answer?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 232227,
      "author_name": "zhinanwang",
      "author_url": "",
      "post_date": "10/17/2017 06:35:12",
      "content": "<p>I think it's future information because it improve my OFFLINE model better than ONLINE.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 237881,
      "author_name": "louis925",
      "author_url": "",
      "post_date": "10/31/2017 07:40:39",
      "content": "<p>I think it is future data. For example, you can see the first item in submission msno <code>'ugx0CjOMzazClkFzU2xasmDZaoIqOUAZPsH1q0teWCg='</code>. In the transactions.csv, the last transaction is on 20170228 with <code>membership_expire_date = 20170330</code> and <code>payment_plan_days = 30</code>. However, in <code>members.csv</code>, the <code>expiration_date</code> is <code>20170822</code>. From the transaction, you cannot know if the user has subscribe the service till 201708. And, if you know the user has expiration date on 201708, there is a very high chance that the user is not churn.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "223942": "",
    "224056": "I guess this should be equal to the last expiration date updated in the transactions until end of Feb... in such case it won't be forward looking.",
    "224111": "In my view, I think it's future information, I don't use it",
    "228793": "No answer?",
    "232227": "I think it's future information because it improve my OFFLINE model better than ONLINE.",
    "237881": "I think it is future data. For example, you can see the first item in submission msno `'ugx0CjOMzazClkFzU2xasmDZaoIqOUAZPsH1q0teWCg='`. In the transactions.csv, the last transaction is on 20170228 with `membership_expire_date = 20170330` and `payment_plan_days = 30`. However, in `members.csv`, the `expiration_date` is `20170822`. From the transaction, you cannot know if the user has subscribe the service till 201708. And, if you know the user has expiration date on 201708, there is a very high chance that the user is not churn."
  },
  "source": "meta"
}