{
  "id": 120474,
  "title": "How to proceed with single model",
  "url": "/competitions/ashrae-energy-prediction/discussion/120474",
  "author_name": "",
  "post_date": "2019-12-06T09:59:57.786889Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I tried with 20 different models by making some buildings and there meters one hot encoded and fed them in LSTM to study the trend but it seems to be wrong way. Please give a hint to start with single model</p>",
  "messages": [
    {
      "id": "688978",
      "postDate": "12/06/2019 09:59:57",
      "content": "<p>I tried with 20 different models by making some buildings and there meters one hot encoded and fed them in LSTM to study the trend but it seems to be wrong way. Please give a hint to start with single model</p>",
      "rawMarkdown": "I tried with 20 different models by making some buildings and there meters one hot encoded and fed them in LSTM to study the trend but it seems to be wrong way. Please give a hint to start with single model",
      "votes": null
    },
    {
      "id": "689040",
      "postDate": "12/06/2019 11:33:16",
      "content": "<p>Don't use LSTM and any time series model.\nYou have to forecast over 2 years with 1 years of data... quite difficult.\nUse tabular model : lgbm, xgboost, nn ,...</p>",
      "rawMarkdown": "Don't use LSTM and any time series model.\nYou have to forecast over 2 years with 1 years of data... quite difficult.\nUse tabular model : lgbm, xgboost, nn ,...",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 689040,
      "author_name": "stenford23",
      "author_url": "",
      "post_date": "12/06/2019 11:33:16",
      "content": "<p>Don't use LSTM and any time series model.\nYou have to forecast over 2 years with 1 years of data... quite difficult.\nUse tabular model : lgbm, xgboost, nn ,...</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "688978": "I tried with 20 different models by making some buildings and there meters one hot encoded and fed them in LSTM to study the trend but it seems to be wrong way. Please give a hint to start with single model",
    "689040": "Don't use LSTM and any time series model.\nYou have to forecast over 2 years with 1 years of data... quite difficult.\nUse tabular model : lgbm, xgboost, nn ,..."
  },
  "source": "meta"
}