{
  "id": 488050,
  "title": "Relationship To Client (Trust 415)",
  "url": "/competitions/home-credit-credit-risk-model-stability/discussion/488050",
  "author_name": "",
  "post_date": "2024-03-31T23:22:34.805434900Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>So there are two relationship columns in train_person_1:</p>\n<ul>\n<li><strong>relationshiptoclient_415T</strong></li>\n<li><strong>relationshiptoclient_642T</strong></li>\n</ul>\n<p><strong>relationshiptoclient_642T</strong> has cases where there are two spouses so I would say that is not really the case</p>\n<p>Also FYI <strong>persontype_792L</strong> goes to <strong>relationshiptoclient_415T</strong> with 1 being yourself, 4 being spouse, and 5 being others. Something to do with spouses not being dependents probably.</p>\n<p><strong>persontype_1072L</strong> goes with <strong>relationshiptoclient_642T</strong> with 1 also filling None places, but sometimes it is nan (all with num_group1 being 1) probably just weird labeling? </p>\n<p>And final pretty much proving it is <strong>personindex_1023L</strong> which directly goes with <strong>relationshiptoclient_415T</strong> and there is no other personindex column</p>",
  "messages": [
    {
      "id": "2726028",
      "postDate": "03/31/2024 23:22:34",
      "content": "<p>So there are two relationship columns in train_person_1:</p>\n<ul>\n<li><strong>relationshiptoclient_415T</strong></li>\n<li><strong>relationshiptoclient_642T</strong></li>\n</ul>\n<p><strong>relationshiptoclient_642T</strong> has cases where there are two spouses so I would say that is not really the case</p>\n<p>Also FYI <strong>persontype_792L</strong> goes to <strong>relationshiptoclient_415T</strong> with 1 being yourself, 4 being spouse, and 5 being others. Something to do with spouses not being dependents probably.</p>\n<p><strong>persontype_1072L</strong> goes with <strong>relationshiptoclient_642T</strong> with 1 also filling None places, but sometimes it is nan (all with num_group1 being 1) probably just weird labeling? </p>\n<p>And final pretty much proving it is <strong>personindex_1023L</strong> which directly goes with <strong>relationshiptoclient_415T</strong> and there is no other personindex column</p>",
      "rawMarkdown": "So there are two relationship columns in train_person_1:\n- **relationshiptoclient_415T**\n- **relationshiptoclient_642T**\n\n**relationshiptoclient_642T** has cases where there are two spouses so I would say that is not really the case\n\nAlso FYI **persontype_792L** goes to **relationshiptoclient_415T** with 1 being yourself, 4 being spouse, and 5 being others. Something to do with spouses not being dependents probably.\n\n**persontype_1072L** goes with **relationshiptoclient_642T** with 1 also filling None places, but sometimes it is nan (all with num_group1 being 1) probably just weird labeling? \n\nAnd final pretty much proving it is **personindex_1023L** which directly goes with **relationshiptoclient_415T** and there is no other personindex column",
      "votes": null
    },
    {
      "id": "2726234",
      "postDate": "04/01/2024 04:16:50",
      "content": "<p>very nice find!</p>",
      "rawMarkdown": "very nice find!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2726234,
      "author_name": "shreyas9181",
      "author_url": "",
      "post_date": "04/01/2024 04:16:50",
      "content": "<p>very nice find!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2726028": "So there are two relationship columns in train_person_1:\n- **relationshiptoclient_415T**\n- **relationshiptoclient_642T**\n\n**relationshiptoclient_642T** has cases where there are two spouses so I would say that is not really the case\n\nAlso FYI **persontype_792L** goes to **relationshiptoclient_415T** with 1 being yourself, 4 being spouse, and 5 being others. Something to do with spouses not being dependents probably.\n\n**persontype_1072L** goes with **relationshiptoclient_642T** with 1 also filling None places, but sometimes it is nan (all with num_group1 being 1) probably just weird labeling? \n\nAnd final pretty much proving it is **personindex_1023L** which directly goes with **relationshiptoclient_415T** and there is no other personindex column",
    "2726234": "very nice find!"
  },
  "source": "meta"
}