{
  "id": 294799,
  "title": "Submitting Lagged Features via API",
  "url": "/competitions/g-research-crypto-forecasting/discussion/294799",
  "author_name": "",
  "post_date": "2021-12-12T22:11:16.888955300Z",
  "votes": 9,
  "comment_count": 1,
  "views": 0,
  "content": "<p>There have been discussions around how to submit models with lagged features via the API.</p>\n<p>I have made an attempt in the notebook linked below. I record the output from the api in a 'history' dataframe which can be accessed when making features. </p>\n<p>The submission runs in 7 hours, although there is plenty of room for speeding up the logic within the api loop. I may update the code in the near future with a sped up version.</p>\n<p><a href=\"url\" target=\"_blank\">https://www.kaggle.com/tomforbes/gresearch-submitting-lagged-features-via-api</a></p>",
  "messages": [
    {
      "id": "1615901",
      "postDate": "12/12/2021 22:11:16",
      "content": "<p>There have been discussions around how to submit models with lagged features via the API.</p>\n<p>I have made an attempt in the notebook linked below. I record the output from the api in a 'history' dataframe which can be accessed when making features. </p>\n<p>The submission runs in 7 hours, although there is plenty of room for speeding up the logic within the api loop. I may update the code in the near future with a sped up version.</p>\n<p><a href=\"url\" target=\"_blank\">https://www.kaggle.com/tomforbes/gresearch-submitting-lagged-features-via-api</a></p>",
      "rawMarkdown": "There have been discussions around how to submit models with lagged features via the API.\n\nI have made an attempt in the notebook linked below. I record the output from the api in a 'history' dataframe which can be accessed when making features. \n\nThe submission runs in 7 hours, although there is plenty of room for speeding up the logic within the api loop. I may update the code in the near future with a sped up version.\n\n[https://www.kaggle.com/tomforbes/gresearch-submitting-lagged-features-via-api](url)",
      "votes": null
    },
    {
      "id": "1616793",
      "postDate": "12/13/2021 16:10:32",
      "content": "<p>great work</p>",
      "rawMarkdown": "great work",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1616793,
      "author_name": "yus002",
      "author_url": "",
      "post_date": "12/13/2021 16:10:32",
      "content": "<p>great work</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1615901": "There have been discussions around how to submit models with lagged features via the API.\n\nI have made an attempt in the notebook linked below. I record the output from the api in a 'history' dataframe which can be accessed when making features. \n\nThe submission runs in 7 hours, although there is plenty of room for speeding up the logic within the api loop. I may update the code in the near future with a sped up version.\n\n[https://www.kaggle.com/tomforbes/gresearch-submitting-lagged-features-via-api](url)",
    "1616793": "great work"
  },
  "source": "meta"
}