{
  "id": 112881,
  "title": "Is this really prediction?",
  "url": "/competitions/ashrae-energy-prediction/discussion/112881",
  "author_name": "",
  "post_date": "2019-10-16T00:48:12.792045600Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Looks like the competition assumes future weather values are available. \nSounds more like an estimation problem (although they are both similar, there are differences specially in terms of pragmatic solutions). </p>",
  "messages": [
    {
      "id": "649962",
      "postDate": "10/16/2019 00:48:12",
      "content": "<p>Looks like the competition assumes future weather values are available. \nSounds more like an estimation problem (although they are both similar, there are differences specially in terms of pragmatic solutions). </p>",
      "rawMarkdown": "Looks like the competition assumes future weather values are available. \nSounds more like an estimation problem (although they are both similar, there are differences specially in terms of pragmatic solutions).",
      "votes": null
    },
    {
      "id": "650594",
      "postDate": "10/16/2019 14:31:16",
      "content": "<p>This is about modeling a counterfactual, not the future, so weather data really is available for these sorts of problems.</p>",
      "rawMarkdown": "This is about modeling a counterfactual, not the future, so weather data really is available for these sorts of problems.",
      "votes": null
    },
    {
      "id": "650943",
      "postDate": "10/16/2019 21:32:35",
      "content": "<p>Thanks, that helps. I guess it makes sense (although the title is maybe misleading :) ) . \nSo a practical scenario example would be:</p>\n\n<p>1) building gets created  in 2016\n3) Calibrate model with kW &amp; weather data from 2016-2018 \n4) Upgrade building in 2019 and measure its kW \n5) Run Model from 3) with weather data from 2019\n6) Perform A/B testing of kW usages between steps 4 and 5 </p>",
      "rawMarkdown": "Thanks, that helps. I guess it makes sense (although the title is maybe misleading :) ) . \nSo a practical scenario example would be:\n\n1) building gets created  in 2016\n3) Calibrate model with kW &amp; weather data from 2016-2018 \n4) Upgrade building in 2019 and measure its kW \n5) Run Model from 3) with weather data from 2019\n6) Perform A/B testing of kW usages between steps 4 and 5",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 650594,
      "author_name": "sohier",
      "author_url": "",
      "post_date": "10/16/2019 14:31:16",
      "content": "<p>This is about modeling a counterfactual, not the future, so weather data really is available for these sorts of problems.</p>",
      "votes": null,
      "replies": [
        {
          "id": 650943,
          "author_name": "ptolemy",
          "author_url": "",
          "post_date": "10/16/2019 21:32:35",
          "content": "<p>Thanks, that helps. I guess it makes sense (although the title is maybe misleading :) ) . \nSo a practical scenario example would be:</p>\n\n<p>1) building gets created  in 2016\n3) Calibrate model with kW &amp; weather data from 2016-2018 \n4) Upgrade building in 2019 and measure its kW \n5) Run Model from 3) with weather data from 2019\n6) Perform A/B testing of kW usages between steps 4 and 5 </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "649962": "Looks like the competition assumes future weather values are available. \nSounds more like an estimation problem (although they are both similar, there are differences specially in terms of pragmatic solutions).",
    "650594": "This is about modeling a counterfactual, not the future, so weather data really is available for these sorts of problems.",
    "650943": "Thanks, that helps. I guess it makes sense (although the title is maybe misleading :) ) . \nSo a practical scenario example would be:\n\n1) building gets created  in 2016\n3) Calibrate model with kW &amp; weather data from 2016-2018 \n4) Upgrade building in 2019 and measure its kW \n5) Run Model from 3) with weather data from 2019\n6) Perform A/B testing of kW usages between steps 4 and 5"
  },
  "source": "meta"
}