{
  "id": 508078,
  "title": "Does the same person appear in the data?",
  "url": "/competitions/home-credit-credit-risk-model-stability/discussion/508078",
  "author_name": "Yohei Tomio",
  "post_date": "2024-05-28T07:07:32.791000",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>In this data, I believe there might be rows (case_id) representing defaults at different time periods for the same person.<br>\nIn the previous competition, which is the predecessor to this one,  “Data Property“ were found. More information can be found here: <a href=\"https://www.kaggle.com/code/titericz/giba-post-processing-user-id-boost\" target=\"_blank\">Home Credit Default Risk</a>.<br>\nI noticed that when the values in the columns [\"max_birth_259D\", \"max_dateofcredstart_181D\", \"max_empl_employedfrom_271D\", \"max_sex_738L\"] match, the values in the other columns also tend to be very close.</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7583266%2F4a9a9d8e709d85b788f1d3b0d271c931%2Fscreen-shot%20-kaggle-home-credit.png?generation=1716879823198124&amp;alt=media\"></p>\n<p>I experimented with this information to improve the  accuracy, but it didn't yield the desired results. Perhaps the assumption that the same person is included is incorrect.<br>\nHas anyone else found such data properties?</p>",
  "messages": [
    {
      "id": 2840656,
      "postDate": "2024-05-28T07:31:17.970Z",
      "content": "<p>I believe so, I have looked for same customers if their date columns differences matches, such as:<br>\nfirstdatedue_489D - birth_259D and<br>\nfirstclxcampaign_1125D - birth_259D</p>",
      "rawMarkdown": "I believe so, I have looked for same customers if their date columns differences matches, such as:\nfirstdatedue_489D - birth_259D and\nfirstclxcampaign_1125D - birth_259D",
      "votes": 1
    },
    {
      "id": 2840624,
      "postDate": "2024-05-28T07:07:32.793Z",
      "content": "<p>In this data, I believe there might be rows (case_id) representing defaults at different time periods for the same person.<br>\nIn the previous competition, which is the predecessor to this one,  “Data Property“ were found. More information can be found here: <a href=\"https://www.kaggle.com/code/titericz/giba-post-processing-user-id-boost\" target=\"_blank\">Home Credit Default Risk</a>.<br>\nI noticed that when the values in the columns [\"max_birth_259D\", \"max_dateofcredstart_181D\", \"max_empl_employedfrom_271D\", \"max_sex_738L\"] match, the values in the other columns also tend to be very close.</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7583266%2F4a9a9d8e709d85b788f1d3b0d271c931%2Fscreen-shot%20-kaggle-home-credit.png?generation=1716879823198124&amp;alt=media\"></p>\n<p>I experimented with this information to improve the  accuracy, but it didn't yield the desired results. Perhaps the assumption that the same person is included is incorrect.<br>\nHas anyone else found such data properties?</p>",
      "rawMarkdown": "In this data, I believe there might be rows (case_id) representing defaults at different time periods for the same person.\nIn the previous competition, which is the predecessor to this one,  “Data Property“ were found. More information can be found here: [Home Credit Default Risk](https://www.kaggle.com/code/titericz/giba-post-processing-user-id-boost).\nI noticed that when the values in the columns [\"max_birth_259D\", \"max_dateofcredstart_181D\", \"max_empl_employedfrom_271D\", \"max_sex_738L\"] match, the values in the other columns also tend to be very close.\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7583266%2F4a9a9d8e709d85b788f1d3b0d271c931%2Fscreen-shot%20-kaggle-home-credit.png?generation=1716879823198124&alt=media)\n\n\nI experimented with this information to improve the  accuracy, but it didn't yield the desired results. Perhaps the assumption that the same person is included is incorrect.\nHas anyone else found such data properties?\n"
    },
    {
      "id": 2840711,
      "postDate": "2024-05-28T08:17:22.820Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 2840656,
      "author_name": "Andrey Chankin",
      "author_url": "",
      "post_date": "2024-05-28T07:31:17.970000",
      "content": "<p>I believe so, I have looked for same customers if their date columns differences matches, such as:<br>\nfirstdatedue_489D - birth_259D and<br>\nfirstclxcampaign_1125D - birth_259D</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 2840711,
      "author_name": "",
      "author_url": "",
      "post_date": "2024-05-28T08:17:22.820000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2840656": "I believe so, I have looked for same customers if their date columns differences matches, such as:\nfirstdatedue_489D - birth_259D and\nfirstclxcampaign_1125D - birth_259D",
    "2840624": "In this data, I believe there might be rows (case_id) representing defaults at different time periods for the same person.\nIn the previous competition, which is the predecessor to this one,  “Data Property“ were found. More information can be found here: [Home Credit Default Risk](https://www.kaggle.com/code/titericz/giba-post-processing-user-id-boost).\nI noticed that when the values in the columns [\"max_birth_259D\", \"max_dateofcredstart_181D\", \"max_empl_employedfrom_271D\", \"max_sex_738L\"] match, the values in the other columns also tend to be very close.\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7583266%2F4a9a9d8e709d85b788f1d3b0d271c931%2Fscreen-shot%20-kaggle-home-credit.png?generation=1716879823198124&alt=media)\n\n\nI experimented with this information to improve the  accuracy, but it didn't yield the desired results. Perhaps the assumption that the same person is included is incorrect.\nHas anyone else found such data properties?\n",
    "2840711": ""
  }
}