{
  "id": 335158,
  "title": "Doubt on 18 months window",
  "url": "/competitions/amex-default-prediction/discussion/335158",
  "author_name": "Jayesh Gokhale",
  "post_date": "2022-07-05T03:56:09.014000",
  "votes": 17,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello Everyone,</p>\n<p>I have some doubts on the problem framing.</p>\n<p>\"The target binary variable is calculated by observing 18 months performance window after the latest credit card statement, and if the customer does not pay due amount in 120 days after their latest statement date it is considered a default event.\"</p>\n<p>Now what I understood is that if the customer has not paid due amount in 120 days after statement, it is considered default. In the data we have only one record per customer for target variable. Hence I would assume that the default or no-default would be based on the latest statement (March 2018 for most cases). This has been provided to us based on the information that bank has whether the due as on March 2018 was paid by July 2018 or not. Please let me know if my assumptions is incorrect.</p>\n<p>Where does the 18 months window come into the picture if the call on default/no-default is made after 120 days? If after 120 days, the customer has not paid, it is marked as default and if customer has paid, in any case bank recovered its money. I understand new statements will be generated in 120 day period, but still not clear on the concept of 18 months and where does this come into the picture (especially for modelling)?</p>\n<p>Thanks</p>",
  "messages": [
    {
      "id": 1843628,
      "postDate": "2022-07-05T03:56:09.013Z",
      "content": "<p>Hello Everyone,</p>\n<p>I have some doubts on the problem framing.</p>\n<p>\"The target binary variable is calculated by observing 18 months performance window after the latest credit card statement, and if the customer does not pay due amount in 120 days after their latest statement date it is considered a default event.\"</p>\n<p>Now what I understood is that if the customer has not paid due amount in 120 days after statement, it is considered default. In the data we have only one record per customer for target variable. Hence I would assume that the default or no-default would be based on the latest statement (March 2018 for most cases). This has been provided to us based on the information that bank has whether the due as on March 2018 was paid by July 2018 or not. Please let me know if my assumptions is incorrect.</p>\n<p>Where does the 18 months window come into the picture if the call on default/no-default is made after 120 days? If after 120 days, the customer has not paid, it is marked as default and if customer has paid, in any case bank recovered its money. I understand new statements will be generated in 120 day period, but still not clear on the concept of 18 months and where does this come into the picture (especially for modelling)?</p>\n<p>Thanks</p>",
      "rawMarkdown": "Hello Everyone,\n\nI have some doubts on the problem framing.\n\n\"The target binary variable is calculated by observing 18 months performance window after the latest credit card statement, and if the customer does not pay due amount in 120 days after their latest statement date it is considered a default event.\"\n\nNow what I understood is that if the customer has not paid due amount in 120 days after statement, it is considered default. In the data we have only one record per customer for target variable. Hence I would assume that the default or no-default would be based on the latest statement (March 2018 for most cases). This has been provided to us based on the information that bank has whether the due as on March 2018 was paid by July 2018 or not. Please let me know if my assumptions is incorrect.\n\nWhere does the 18 months window come into the picture if the call on default/no-default is made after 120 days? If after 120 days, the customer has not paid, it is marked as default and if customer has paid, in any case bank recovered its money. I understand new statements will be generated in 120 day period, but still not clear on the concept of 18 months and where does this come into the picture (especially for modelling)?\n\nThanks",
      "votes": 16
    },
    {
      "id": 1844649,
      "postDate": "2022-07-05T17:13:52.263Z",
      "content": "<p>Thanks for asking this, I am also unclear on this.</p>",
      "rawMarkdown": "Thanks for asking this, I am also unclear on this.",
      "votes": 1
    },
    {
      "id": 1844950,
      "postDate": "2022-07-06T00:00:25.197Z",
      "rawMarkdown": "",
      "votes": 2,
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 1844649,
      "author_name": "Darien Schettler",
      "author_url": "",
      "post_date": "2022-07-05T17:13:52.263000",
      "content": "<p>Thanks for asking this, I am also unclear on this.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 1844950,
      "author_name": "",
      "author_url": "",
      "post_date": "2022-07-06T00:00:25.197000",
      "content": "",
      "votes": 2,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1843628": "Hello Everyone,\n\nI have some doubts on the problem framing.\n\n\"The target binary variable is calculated by observing 18 months performance window after the latest credit card statement, and if the customer does not pay due amount in 120 days after their latest statement date it is considered a default event.\"\n\nNow what I understood is that if the customer has not paid due amount in 120 days after statement, it is considered default. In the data we have only one record per customer for target variable. Hence I would assume that the default or no-default would be based on the latest statement (March 2018 for most cases). This has been provided to us based on the information that bank has whether the due as on March 2018 was paid by July 2018 or not. Please let me know if my assumptions is incorrect.\n\nWhere does the 18 months window come into the picture if the call on default/no-default is made after 120 days? If after 120 days, the customer has not paid, it is marked as default and if customer has paid, in any case bank recovered its money. I understand new statements will be generated in 120 day period, but still not clear on the concept of 18 months and where does this come into the picture (especially for modelling)?\n\nThanks",
    "1844649": "Thanks for asking this, I am also unclear on this.",
    "1844950": ""
  }
}