{
  "id": 369772,
  "title": "Is the timestamp sorted?",
  "url": "/competitions/otto-recommender-system/discussion/369772",
  "author_name": "",
  "post_date": "2022-12-01T10:47:28.872130600Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Sorry if already answered, but the timestamp is sorted right?</p>",
  "messages": [
    {
      "id": "2051331",
      "postDate": "12/01/2022 10:47:28",
      "content": "<p>Sorry if already answered, but the timestamp is sorted right?</p>",
      "rawMarkdown": "Sorry if already answered, but the timestamp is sorted right?",
      "votes": null
    },
    {
      "id": "2051471",
      "postDate": "12/01/2022 12:54:19",
      "content": "<p>The answer should be yes 🙂 From what I understand, <code>pandas</code> keeps the ordering across most (all?) operations, but other frameworks that prioritize speed of execution (mostly via parallelizing operations) do not.</p>\n<p>If in doubt, you can always verify this 🙂 <code>dataset.sort_values(by='ts') == dataset.ts</code> should give you the answer whether the ds is sorted or not 🙂 But, for some things, like training ranking model, you want the dataset sorted by ['session', 'ts'].</p>\n<p>Anyhow, hope this can be of help 🙂</p>",
      "rawMarkdown": "The answer should be yes 🙂 From what I understand, `pandas` keeps the ordering across most (all?) operations, but other frameworks that prioritize speed of execution (mostly via parallelizing operations) do not.\n\nIf in doubt, you can always verify this 🙂 `dataset.sort_values(by='ts') == dataset.ts` should give you the answer whether the ds is sorted or not 🙂 But, for some things, like training ranking model, you want the dataset sorted by ['session', 'ts'].\n\nAnyhow, hope this can be of help 🙂",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2051471,
      "author_name": "radek1",
      "author_url": "",
      "post_date": "12/01/2022 12:54:19",
      "content": "<p>The answer should be yes 🙂 From what I understand, <code>pandas</code> keeps the ordering across most (all?) operations, but other frameworks that prioritize speed of execution (mostly via parallelizing operations) do not.</p>\n<p>If in doubt, you can always verify this 🙂 <code>dataset.sort_values(by='ts') == dataset.ts</code> should give you the answer whether the ds is sorted or not 🙂 But, for some things, like training ranking model, you want the dataset sorted by ['session', 'ts'].</p>\n<p>Anyhow, hope this can be of help 🙂</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2051331": "Sorry if already answered, but the timestamp is sorted right?",
    "2051471": "The answer should be yes 🙂 From what I understand, `pandas` keeps the ordering across most (all?) operations, but other frameworks that prioritize speed of execution (mostly via parallelizing operations) do not.\n\nIf in doubt, you can always verify this 🙂 `dataset.sort_values(by='ts') == dataset.ts` should give you the answer whether the ds is sorted or not 🙂 But, for some things, like training ranking model, you want the dataset sorted by ['session', 'ts'].\n\nAnyhow, hope this can be of help 🙂"
  },
  "source": "meta"
}