{
  "id": 547370,
  "title": "About variables \"date_id\" and \"time_id\"",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/547370",
  "author_name": "",
  "post_date": "2024-11-21T07:15:53.685014600Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello everyone. I am stuck with the \"date_id\" and \"time_id\" variables. Any ideas on how to convert them into regular time and date stamps?</p>",
  "messages": [
    {
      "id": "3051325",
      "postDate": "11/21/2024 07:15:53",
      "content": "<p>Hello everyone. I am stuck with the \"date_id\" and \"time_id\" variables. Any ideas on how to convert them into regular time and date stamps?</p>",
      "rawMarkdown": "Hello everyone. I am stuck with the \"date_id\" and \"time_id\" variables. Any ideas on how to convert them into regular time and date stamps?",
      "votes": null
    },
    {
      "id": "3051435",
      "postDate": "11/21/2024 09:35:11",
      "content": "<p>You can't, given they're anonimized. They explained you might have 1 min difference betweent two time_ids and then 3 min between other two time_ids. For dates we might be looking at decade-old data or recent data, we don't know</p>",
      "rawMarkdown": "You can't, given they're anonimized. They explained you might have 1 min difference betweent two time_ids and then 3 min between other two time_ids. For dates we might be looking at decade-old data or recent data, we don't know",
      "votes": null
    },
    {
      "id": "3054317",
      "postDate": "11/24/2024 14:56:59",
      "content": "<p>As mentioned in the Overview of the competition, the two mentioned features are <strong>ordinal</strong> so you can compare two rows in time by first comparing their  **date_id  **column, and then comapre <strong>time_id</strong> (only for an equal <strong>date_id</strong>) .</p>\n<p>So what you can do is to have a **combined feature = 1000 * date_id + time_id ** and then sorting your data on this feature will give you chronological ordered data.</p>",
      "rawMarkdown": "As mentioned in the Overview of the competition, the two mentioned features are **ordinal** so you can compare two rows in time by first comparing their  **date_id  **column, and then comapre **time_id** (only for an equal **date_id**) .\n\nSo what you can do is to have a **combined feature = 1000 * date_id + time_id ** and then sorting your data on this feature will give you chronological ordered data.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3051435,
      "author_name": "natanlabarrere",
      "author_url": "",
      "post_date": "11/21/2024 09:35:11",
      "content": "<p>You can't, given they're anonimized. They explained you might have 1 min difference betweent two time_ids and then 3 min between other two time_ids. For dates we might be looking at decade-old data or recent data, we don't know</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3054317,
      "author_name": "ayoubchouikha",
      "author_url": "",
      "post_date": "11/24/2024 14:56:59",
      "content": "<p>As mentioned in the Overview of the competition, the two mentioned features are <strong>ordinal</strong> so you can compare two rows in time by first comparing their  **date_id  **column, and then comapre <strong>time_id</strong> (only for an equal <strong>date_id</strong>) .</p>\n<p>So what you can do is to have a **combined feature = 1000 * date_id + time_id ** and then sorting your data on this feature will give you chronological ordered data.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3051325": "Hello everyone. I am stuck with the \"date_id\" and \"time_id\" variables. Any ideas on how to convert them into regular time and date stamps?",
    "3051435": "You can't, given they're anonimized. They explained you might have 1 min difference betweent two time_ids and then 3 min between other two time_ids. For dates we might be looking at decade-old data or recent data, we don't know",
    "3054317": "As mentioned in the Overview of the competition, the two mentioned features are **ordinal** so you can compare two rows in time by first comparing their  **date_id  **column, and then comapre **time_id** (only for an equal **date_id**) .\n\nSo what you can do is to have a **combined feature = 1000 * date_id + time_id ** and then sorting your data on this feature will give you chronological ordered data."
  },
  "source": "meta"
}