{
  "id": 117355,
  "title": "Using local machine",
  "url": "/competitions/ashrae-energy-prediction/discussion/117355",
  "author_name": "",
  "post_date": "2019-11-14T21:00:49.954136900Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>This competition has a lot of memory constraints and my kernel crashes while using join, drop, concat etc. on the dataframes, especially on the Test data. However I do not face any issues on my own computer. Is there a way I can generate the submission excel and inject it directly into the kernel? If not, is there a way I can make my computer do the heavy-lifting and use the intermediate results in my kernel?\nI am new to ML and this is the largest dataset I have faced till now.</p>",
  "messages": [
    {
      "id": "673331",
      "postDate": "11/14/2019 21:00:49",
      "content": "<p>This competition has a lot of memory constraints and my kernel crashes while using join, drop, concat etc. on the dataframes, especially on the Test data. However I do not face any issues on my own computer. Is there a way I can generate the submission excel and inject it directly into the kernel? If not, is there a way I can make my computer do the heavy-lifting and use the intermediate results in my kernel?\nI am new to ML and this is the largest dataset I have faced till now.</p>",
      "rawMarkdown": "This competition has a lot of memory constraints and my kernel crashes while using join, drop, concat etc. on the dataframes, especially on the Test data. However I do not face any issues on my own computer. Is there a way I can generate the submission excel and inject it directly into the kernel? If not, is there a way I can make my computer do the heavy-lifting and use the intermediate results in my kernel?\nI am new to ML and this is the largest dataset I have faced till now.",
      "votes": null
    },
    {
      "id": "673350",
      "postDate": "11/14/2019 21:46:02",
      "content": "<p>Use reduct_memory_reduction functions (search on kaggle, it's a function which reduce the memory consumption by downcasting the type of variables) on building dataset, weather and train before join.</p>",
      "rawMarkdown": "Use reduct_memory_reduction functions (search on kaggle, it's a function which reduce the memory consumption by downcasting the type of variables) on building dataset, weather and train before join.",
      "votes": null
    },
    {
      "id": "673633",
      "postDate": "11/15/2019 08:52:03",
      "content": "<p>You may check the feather format to load the data as well - solves a lot of memory issues.  </p>",
      "rawMarkdown": "You may check the feather format to load the data as well - solves a lot of memory issues.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 673350,
      "author_name": "stenford23",
      "author_url": "",
      "post_date": "11/14/2019 21:46:02",
      "content": "<p>Use reduct_memory_reduction functions (search on kaggle, it's a function which reduce the memory consumption by downcasting the type of variables) on building dataset, weather and train before join.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 673633,
      "author_name": "imeintanis",
      "author_url": "",
      "post_date": "11/15/2019 08:52:03",
      "content": "<p>You may check the feather format to load the data as well - solves a lot of memory issues.  </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "673331": "This competition has a lot of memory constraints and my kernel crashes while using join, drop, concat etc. on the dataframes, especially on the Test data. However I do not face any issues on my own computer. Is there a way I can generate the submission excel and inject it directly into the kernel? If not, is there a way I can make my computer do the heavy-lifting and use the intermediate results in my kernel?\nI am new to ML and this is the largest dataset I have faced till now.",
    "673350": "Use reduct_memory_reduction functions (search on kaggle, it's a function which reduce the memory consumption by downcasting the type of variables) on building dataset, weather and train before join.",
    "673633": "You may check the feather format to load the data as well - solves a lot of memory issues."
  },
  "source": "meta"
}