{
  "id": 115174,
  "title": "general question about the data",
  "url": "/competitions/ashrae-energy-prediction/discussion/115174",
  "author_name": "",
  "post_date": "2019-10-31T19:56:33.473190600Z",
  "votes": null,
  "comment_count": 7,
  "views": 0,
  "content": "<p>1st timer here on Kaggle.. Could someone give me a tip if the data needs to be broken down for each building? IE, each building would have one dependent variable (meter reading) and many independent variables (weather) So one algorithm per building_id?</p>",
  "messages": [
    {
      "id": "662649",
      "postDate": "10/31/2019 19:56:33",
      "content": "<p>1st timer here on Kaggle.. Could someone give me a tip if the data needs to be broken down for each building? IE, each building would have one dependent variable (meter reading) and many independent variables (weather) So one algorithm per building_id?</p>",
      "rawMarkdown": "1st timer here on Kaggle.. Could someone give me a tip if the data needs to be broken down for each building? IE, each building would have one dependent variable (meter reading) and many independent variables (weather) So one algorithm per building_id?",
      "votes": null
    },
    {
      "id": "662657",
      "postDate": "10/31/2019 20:23:51",
      "content": "<p>There is no right or wrong answer, you just have to explore :-). <code>meter_reading</code> is indeed your dependent variable, but as far as how you get to it...</p>",
      "rawMarkdown": "There is no right or wrong answer, you just have to explore :-). `meter_reading` is indeed your dependent variable, but as far as how you get to it...",
      "votes": null
    },
    {
      "id": "662665",
      "postDate": "10/31/2019 20:36:06",
      "content": "<p>Thanks. Is it legitimate per kaggle standards to break down all the data into individual CSVs and join them together again?</p>",
      "rawMarkdown": "Thanks. Is it legitimate per kaggle standards to break down all the data into individual CSVs and join them together again?",
      "votes": null
    },
    {
      "id": "662673",
      "postDate": "10/31/2019 20:52:44",
      "content": "<p>For each meter is better than for each building.at least for me</p>",
      "rawMarkdown": "For each meter is better than for each building.at least for me",
      "votes": null
    },
    {
      "id": "662695",
      "postDate": "10/31/2019 21:40:48",
      "content": "<p>You should watch out for some buildings like 403, 604 as they have only several hundreds of examples in the training set.\nI agree with Oleg. In this case you should try to make your model learn from other similar buildings to predict for example building id 403 and 604.</p>",
      "rawMarkdown": "You should watch out for some buildings like 403, 604 as they have only several hundreds of examples in the training set.\nI agree with Oleg. In this case you should try to make your model learn from other similar buildings to predict for example building id 403 and 604.",
      "votes": null
    },
    {
      "id": "662737",
      "postDate": "10/31/2019 23:48:07",
      "content": "<p>Do you have the best results with deep learning MLP?</p>",
      "rawMarkdown": "Do you have the best results with deep learning MLP?",
      "votes": null
    },
    {
      "id": "662803",
      "postDate": "11/01/2019 02:48:44",
      "content": "<p>There are no Kaggle standards. That stated, I wouldn't recommend doing this. Even if you're hurting for RAM, it'll still be more straightforward to stream the loading of your data into chunks in my opinion.</p>",
      "rawMarkdown": "There are no Kaggle standards. That stated, I wouldn't recommend doing this. Even if you're hurting for RAM, it'll still be more straightforward to stream the loading of your data into chunks in my opinion.",
      "votes": null
    },
    {
      "id": "663260",
      "postDate": "11/01/2019 17:33:56",
      "content": "<p>Hey <a href=\"/authman\">@authman</a>, one quick question. What's the problem of breaking down into CSV's?</p>",
      "rawMarkdown": "Hey @authman, one quick question. What's the problem of breaking down into CSV's?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 662657,
      "author_name": "authman",
      "author_url": "",
      "post_date": "10/31/2019 20:23:51",
      "content": "<p>There is no right or wrong answer, you just have to explore :-). <code>meter_reading</code> is indeed your dependent variable, but as far as how you get to it...</p>",
      "votes": null,
      "replies": [
        {
          "id": 662665,
          "author_name": "henryhub",
          "author_url": "",
          "post_date": "10/31/2019 20:36:06",
          "content": "<p>Thanks. Is it legitimate per kaggle standards to break down all the data into individual CSVs and join them together again?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 662803,
          "author_name": "authman",
          "author_url": "",
          "post_date": "11/01/2019 02:48:44",
          "content": "<p>There are no Kaggle standards. That stated, I wouldn't recommend doing this. Even if you're hurting for RAM, it'll still be more straightforward to stream the loading of your data into chunks in my opinion.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 663260,
          "author_name": "themonologue",
          "author_url": "",
          "post_date": "11/01/2019 17:33:56",
          "content": "<p>Hey <a href=\"/authman\">@authman</a>, one quick question. What's the problem of breaking down into CSV's?</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 662673,
      "author_name": "oleg90777",
      "author_url": "",
      "post_date": "10/31/2019 20:52:44",
      "content": "<p>For each meter is better than for each building.at least for me</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 662695,
      "author_name": "jielu0728",
      "author_url": "",
      "post_date": "10/31/2019 21:40:48",
      "content": "<p>You should watch out for some buildings like 403, 604 as they have only several hundreds of examples in the training set.\nI agree with Oleg. In this case you should try to make your model learn from other similar buildings to predict for example building id 403 and 604.</p>",
      "votes": null,
      "replies": [
        {
          "id": 662737,
          "author_name": "henryhub",
          "author_url": "",
          "post_date": "10/31/2019 23:48:07",
          "content": "<p>Do you have the best results with deep learning MLP?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "662649": "1st timer here on Kaggle.. Could someone give me a tip if the data needs to be broken down for each building? IE, each building would have one dependent variable (meter reading) and many independent variables (weather) So one algorithm per building_id?",
    "662657": "There is no right or wrong answer, you just have to explore :-). `meter_reading` is indeed your dependent variable, but as far as how you get to it...",
    "662665": "Thanks. Is it legitimate per kaggle standards to break down all the data into individual CSVs and join them together again?",
    "662673": "For each meter is better than for each building.at least for me",
    "662695": "You should watch out for some buildings like 403, 604 as they have only several hundreds of examples in the training set.\nI agree with Oleg. In this case you should try to make your model learn from other similar buildings to predict for example building id 403 and 604.",
    "662737": "Do you have the best results with deep learning MLP?",
    "662803": "There are no Kaggle standards. That stated, I wouldn't recommend doing this. Even if you're hurting for RAM, it'll still be more straightforward to stream the loading of your data into chunks in my opinion.",
    "663260": "Hey @authman, one quick question. What's the problem of breaking down into CSV's?"
  },
  "source": "meta"
}