{
  "id": 498267,
  "title": "Preprocessing nearest neighbors to save some time",
  "url": "/competitions/uspto-explainable-ai/discussion/498267",
  "author_name": "Valentin Werner",
  "post_date": "2024-04-27T14:58:35.885000",
  "votes": 3,
  "comment_count": 0,
  "views": 0,
  "content": "<p>From what I understand, the nearest neighbors will be a key file for this competition. But the file is too large for Kaggle memory. Loading in batches is a common procedure for this problem. Even more so, as we only need approx. 200k rows out of the 13M+ patents.</p>\n<p>You can find the notebook and use it is input for your notebooks here: <a href=\"https://www.kaggle.com/code/valentinwerner/save-some-loading-time-tr-idx-subset-of-neighbors/notebook\" target=\"_blank\">https://www.kaggle.com/code/valentinwerner/save-some-loading-time-tr-idx-subset-of-neighbors/notebook</a></p>\n<p>Hope this helps you save some time ☺️</p>",
  "messages": [
    {
      "id": 2779160,
      "postDate": "2024-04-27T14:58:35.887Z",
      "content": "<p>From what I understand, the nearest neighbors will be a key file for this competition. But the file is too large for Kaggle memory. Loading in batches is a common procedure for this problem. Even more so, as we only need approx. 200k rows out of the 13M+ patents.</p>\n<p>You can find the notebook and use it is input for your notebooks here: <a href=\"https://www.kaggle.com/code/valentinwerner/save-some-loading-time-tr-idx-subset-of-neighbors/notebook\" target=\"_blank\">https://www.kaggle.com/code/valentinwerner/save-some-loading-time-tr-idx-subset-of-neighbors/notebook</a></p>\n<p>Hope this helps you save some time ☺️</p>",
      "rawMarkdown": "From what I understand, the nearest neighbors will be a key file for this competition. But the file is too large for Kaggle memory. Loading in batches is a common procedure for this problem. Even more so, as we only need approx. 200k rows out of the 13M+ patents.\n\nYou can find the notebook and use it is input for your notebooks here: https://www.kaggle.com/code/valentinwerner/save-some-loading-time-tr-idx-subset-of-neighbors/notebook\n\nHope this helps you save some time ☺️",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2779160": "From what I understand, the nearest neighbors will be a key file for this competition. But the file is too large for Kaggle memory. Loading in batches is a common procedure for this problem. Even more so, as we only need approx. 200k rows out of the 13M+ patents.\n\nYou can find the notebook and use it is input for your notebooks here: https://www.kaggle.com/code/valentinwerner/save-some-loading-time-tr-idx-subset-of-neighbors/notebook\n\nHope this helps you save some time ☺️"
  }
}