{
  "id": 476876,
  "title": "Are temporal dependencies worth considering?",
  "url": "/competitions/home-credit-credit-risk-model-stability/discussion/476876",
  "author_name": "",
  "post_date": "2024-02-13T20:15:58.856712100Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Regarding the nature of the data (tabular) and the objective of the competition, I have my concerns about the presence of temporal dependencies. While the main task is to predict the probability of loan default, are the historical records associated with the loan decisions a type of time series data?</p>\n<p>Is the nature of the competition purely tabular, or there's an element of time series forecasting?</p>",
  "messages": [
    {
      "id": "2651011",
      "postDate": "02/13/2024 20:15:58",
      "content": "<p>Regarding the nature of the data (tabular) and the objective of the competition, I have my concerns about the presence of temporal dependencies. While the main task is to predict the probability of loan default, are the historical records associated with the loan decisions a type of time series data?</p>\n<p>Is the nature of the competition purely tabular, or there's an element of time series forecasting?</p>",
      "rawMarkdown": "Regarding the nature of the data (tabular) and the objective of the competition, I have my concerns about the presence of temporal dependencies. While the main task is to predict the probability of loan default, are the historical records associated with the loan decisions a type of time series data?\n\nIs the nature of the competition purely tabular, or there's an element of time series forecasting?",
      "votes": null
    },
    {
      "id": "2651806",
      "postDate": "02/14/2024 11:32:59",
      "content": "<p>Yes, the historical records associated with the loan decisions are strong predictors of whether the client is likely to default or not (e.g. history of excessive borrowing, history of missed payments…etc) but not in a time series manner…more like establishing a client behavior pattern.</p>",
      "rawMarkdown": "Yes, the historical records associated with the loan decisions are strong predictors of whether the client is likely to default or not (e.g. history of excessive borrowing, history of missed payments...etc) but not in a time series manner...more like establishing a client behavior pattern.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2651806,
      "author_name": "fadylabib",
      "author_url": "",
      "post_date": "02/14/2024 11:32:59",
      "content": "<p>Yes, the historical records associated with the loan decisions are strong predictors of whether the client is likely to default or not (e.g. history of excessive borrowing, history of missed payments…etc) but not in a time series manner…more like establishing a client behavior pattern.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2651011": "Regarding the nature of the data (tabular) and the objective of the competition, I have my concerns about the presence of temporal dependencies. While the main task is to predict the probability of loan default, are the historical records associated with the loan decisions a type of time series data?\n\nIs the nature of the competition purely tabular, or there's an element of time series forecasting?",
    "2651806": "Yes, the historical records associated with the loan decisions are strong predictors of whether the client is likely to default or not (e.g. history of excessive borrowing, history of missed payments...etc) but not in a time series manner...more like establishing a client behavior pattern."
  },
  "source": "meta"
}