{
  "id": 78687,
  "title": "8738 card_ids with weird dates",
  "url": "/competitions/elo-merchant-category-recommendation/discussion/78687",
  "author_name": "",
  "post_date": "2019-01-26T23:43:42.219477Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>from the transactions datasets, if you pick <strong>card_id == \"C_ID_02e9714314\"</strong>, the first transaction (by ordering purchase_date) was on <strong>2017-01-18 19:45:10</strong>.\nBut that same card has 2017-12 as <strong>first_active_month</strong> on train/test datasets</p>\n\n<p>How can it be \"active since\" 2017-12, but has a first transaction on 2017-01; with -11 months difference?</p>\n\n<p>I've found 8738 card_id with that issue. =/</p>",
  "messages": [
    {
      "id": "461757",
      "postDate": "01/26/2019 23:43:42",
      "content": "<p>from the transactions datasets, if you pick <strong>card_id == \"C_ID_02e9714314\"</strong>, the first transaction (by ordering purchase_date) was on <strong>2017-01-18 19:45:10</strong>.\nBut that same card has 2017-12 as <strong>first_active_month</strong> on train/test datasets</p>\n\n<p>How can it be \"active since\" 2017-12, but has a first transaction on 2017-01; with -11 months difference?</p>\n\n<p>I've found 8738 card_id with that issue. =/</p>",
      "rawMarkdown": "from the transactions datasets, if you pick **card_id == \"C_ID_02e9714314\"**, the first transaction (by ordering purchase_date) was on **2017-01-18 19:45:10**.\nBut that same card has 2017-12 as **first_active_month** on train/test datasets\n\nHow can it be \"active since\" 2017-12, but has a first transaction on 2017-01; with -11 months difference?\n\nI've found 8738 card_id with that issue. =/",
      "votes": null
    },
    {
      "id": "461850",
      "postDate": "01/27/2019 06:40:26",
      "content": "<p>It seems that 5429 card_id's in df_train have first purchase is yearlier than in 'first_active_month' .</p>",
      "rawMarkdown": "It seems that 5429 card_id's in df_train have first purchase is yearlier than in 'first_active_month' .",
      "votes": null
    },
    {
      "id": "462801",
      "postDate": "01/28/2019 23:17:40",
      "content": "<p>I dod not think this will affect too much on your results. I tried to replace those 'firstactivemonth', which are after trainsaction date, by the first trasaction date, but the result did not change at all.</p>",
      "rawMarkdown": "I dod not think this will affect too much on your results. I tried to replace those 'firstactivemonth', which are after trainsaction date, by the first trasaction date, but the result did not change at all.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 461850,
      "author_name": "salamatburj",
      "author_url": "",
      "post_date": "01/27/2019 06:40:26",
      "content": "<p>It seems that 5429 card_id's in df_train have first purchase is yearlier than in 'first_active_month' .</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 462801,
      "author_name": "lftuwujie",
      "author_url": "",
      "post_date": "01/28/2019 23:17:40",
      "content": "<p>I dod not think this will affect too much on your results. I tried to replace those 'firstactivemonth', which are after trainsaction date, by the first trasaction date, but the result did not change at all.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "461757": "from the transactions datasets, if you pick **card_id == \"C_ID_02e9714314\"**, the first transaction (by ordering purchase_date) was on **2017-01-18 19:45:10**.\nBut that same card has 2017-12 as **first_active_month** on train/test datasets\n\nHow can it be \"active since\" 2017-12, but has a first transaction on 2017-01; with -11 months difference?\n\nI've found 8738 card_id with that issue. =/",
    "461850": "It seems that 5429 card_id's in df_train have first purchase is yearlier than in 'first_active_month' .",
    "462801": "I dod not think this will affect too much on your results. I tried to replace those 'firstactivemonth', which are after trainsaction date, by the first trasaction date, but the result did not change at all."
  },
  "source": "meta"
}