{
  "id": 113146,
  "title": "Important Discussions thus far",
  "url": "/competitions/ashrae-energy-prediction/discussion/113146",
  "author_name": "",
  "post_date": "2019-10-17T09:51:46.939076200Z",
  "votes": 10,
  "comment_count": 2,
  "views": 0,
  "content": "<p><strong>Task</strong>\nGiven hourly data of a building, predict future hourly data of building</p>\n\n<p><strong>Evaluation Metric</strong>\nmetric is calculated once using all of the readings from all meter types\nsource: <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113086#latest-650958\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113086#latest-650958</a> \nHow to calculate: <a href=\"https://www.kaggle.com/marknagelberg/rmsle-function\">https://www.kaggle.com/marknagelberg/rmsle-function</a> , <a href=\"https://www.kaggle.com/jpopham91/rmlse-vectorized\">https://www.kaggle.com/jpopham91/rmlse-vectorized</a>, \n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113064#latest-650950\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113064#latest-650950</a> </p>\n\n<p><strong>Domain Knowledge / Getting Started</strong>\n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112858#latest-649854\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112858#latest-649854</a> \n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113080#latest-651137\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113080#latest-651137</a> </p>\n\n<p><strong>Similar Competitions</strong>\n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112855#latest-649925\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112855#latest-649925</a> \n<a href=\"http://drivendata.co/blog/power-laws-forecasting-winners/\">http://drivendata.co/blog/power-laws-forecasting-winners/</a> \n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112958#latest-650382\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112958#latest-650382</a> </p>\n\n<p><strong>Food for Thought</strong>\nTrain only spans 1 year (Test is 2 years; therefore, is Time Series analysis less useful than Regression? But hosts say: “Our shared goals are to uncover (and disseminate to the public) current ‘best practices’ for methods of time series predictions that can be useful in our domain.” <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112985#latest-651212\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112985#latest-651212</a> )\nWhat is a negative precip_depth_1_hr value?\nHow will you handle &gt;30% missing data for floor_count, year_built, cloud_coverage, and precip_depth_1_hr ?\nActual weather vs forecasted weather: <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112887#latest-650545\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112887#latest-650545</a> , <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112884#latest-651115\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112884#latest-651115</a> , <a href=\"https://www.mdpi.com/2504-4990/1/3/56/htm\">https://www.mdpi.com/2504-4990/1/3/56/htm</a> </p>\n\n<p><strong>Feature Importance</strong>\n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112871#latest-650996\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112871#latest-650996</a> </p>\n\n<p><strong>Feature Engineering</strong>\nDays (schools are usually closed during Saturday/Sunday)\nMonths (schools are closed during certain months)\nHolidays (many organizations may close on holidays)</p>\n\n<p><strong>Model Choice</strong>\n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113007#latest-650802\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113007#latest-650802</a> \n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112856#latest-650709\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112856#latest-650709</a> </p>\n\n<p><strong>Dealing with RAM usage</strong>\n<a href=\"https://www.kaggle.com/hamditarek/reducing-memory-size-for-great-energy-predictor\">https://www.kaggle.com/hamditarek/reducing-memory-size-for-great-energy-predictor</a> \n<a href=\"https://www.kaggle.com/yuliagm/how-to-work-with-big-datasets-on-16g-ram-dask\">https://www.kaggle.com/yuliagm/how-to-work-with-big-datasets-on-16g-ram-dask</a> </p>",
  "messages": [
    {
      "id": "651301",
      "postDate": "10/17/2019 09:51:46",
      "content": "<p><strong>Task</strong>\nGiven hourly data of a building, predict future hourly data of building</p>\n\n<p><strong>Evaluation Metric</strong>\nmetric is calculated once using all of the readings from all meter types\nsource: <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113086#latest-650958\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113086#latest-650958</a> \nHow to calculate: <a href=\"https://www.kaggle.com/marknagelberg/rmsle-function\">https://www.kaggle.com/marknagelberg/rmsle-function</a> , <a href=\"https://www.kaggle.com/jpopham91/rmlse-vectorized\">https://www.kaggle.com/jpopham91/rmlse-vectorized</a>, \n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113064#latest-650950\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113064#latest-650950</a> </p>\n\n<p><strong>Domain Knowledge / Getting Started</strong>\n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112858#latest-649854\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112858#latest-649854</a> \n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113080#latest-651137\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113080#latest-651137</a> </p>\n\n<p><strong>Similar Competitions</strong>\n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112855#latest-649925\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112855#latest-649925</a> \n<a href=\"http://drivendata.co/blog/power-laws-forecasting-winners/\">http://drivendata.co/blog/power-laws-forecasting-winners/</a> \n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112958#latest-650382\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112958#latest-650382</a> </p>\n\n<p><strong>Food for Thought</strong>\nTrain only spans 1 year (Test is 2 years; therefore, is Time Series analysis less useful than Regression? But hosts say: “Our shared goals are to uncover (and disseminate to the public) current ‘best practices’ for methods of time series predictions that can be useful in our domain.” <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112985#latest-651212\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112985#latest-651212</a> )\nWhat is a negative precip_depth_1_hr value?\nHow will you handle &gt;30% missing data for floor_count, year_built, cloud_coverage, and precip_depth_1_hr ?\nActual weather vs forecasted weather: <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112887#latest-650545\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112887#latest-650545</a> , <a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112884#latest-651115\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112884#latest-651115</a> , <a href=\"https://www.mdpi.com/2504-4990/1/3/56/htm\">https://www.mdpi.com/2504-4990/1/3/56/htm</a> </p>\n\n<p><strong>Feature Importance</strong>\n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112871#latest-650996\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112871#latest-650996</a> </p>\n\n<p><strong>Feature Engineering</strong>\nDays (schools are usually closed during Saturday/Sunday)\nMonths (schools are closed during certain months)\nHolidays (many organizations may close on holidays)</p>\n\n<p><strong>Model Choice</strong>\n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113007#latest-650802\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113007#latest-650802</a> \n<a href=\"https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112856#latest-650709\">https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112856#latest-650709</a> </p>\n\n<p><strong>Dealing with RAM usage</strong>\n<a href=\"https://www.kaggle.com/hamditarek/reducing-memory-size-for-great-energy-predictor\">https://www.kaggle.com/hamditarek/reducing-memory-size-for-great-energy-predictor</a> \n<a href=\"https://www.kaggle.com/yuliagm/how-to-work-with-big-datasets-on-16g-ram-dask\">https://www.kaggle.com/yuliagm/how-to-work-with-big-datasets-on-16g-ram-dask</a> </p>",
      "rawMarkdown": "**Task**\nGiven hourly data of a building, predict future hourly data of building\n\n**Evaluation Metric**\nmetric is calculated once using all of the readings from all meter types\nsource: https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113086#latest-650958 \nHow to calculate: https://www.kaggle.com/marknagelberg/rmsle-function , https://www.kaggle.com/jpopham91/rmlse-vectorized, \nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/113064#latest-650950 \n\n**Domain Knowledge / Getting Started**\nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112858#latest-649854 \nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/113080#latest-651137 \n\n**Similar Competitions**\nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112855#latest-649925 \nhttp://drivendata.co/blog/power-laws-forecasting-winners/ \nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112958#latest-650382 \n\n**Food for Thought**\nTrain only spans 1 year (Test is 2 years; therefore, is Time Series analysis less useful than Regression? But hosts say: “Our shared goals are to uncover (and disseminate to the public) current ‘best practices’ for methods of time series predictions that can be useful in our domain.” https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112985#latest-651212 )\nWhat is a negative precip_depth_1_hr value?\nHow will you handle &gt;30% missing data for floor_count, year_built, cloud_coverage, and precip_depth_1_hr ?\nActual weather vs forecasted weather: https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112887#latest-650545 , https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112884#latest-651115 , https://www.mdpi.com/2504-4990/1/3/56/htm \n\n**Feature Importance**\nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112871#latest-650996 \n\n**Feature Engineering**\nDays (schools are usually closed during Saturday/Sunday)\nMonths (schools are closed during certain months)\nHolidays (many organizations may close on holidays)\n\n**Model Choice**\nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/113007#latest-650802 \nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112856#latest-650709 \n\n**Dealing with RAM usage**\nhttps://www.kaggle.com/hamditarek/reducing-memory-size-for-great-energy-predictor \nhttps://www.kaggle.com/yuliagm/how-to-work-with-big-datasets-on-16g-ram-dask",
      "votes": null
    },
    {
      "id": "652433",
      "postDate": "10/18/2019 20:54:03",
      "content": "<p>Cool! Thanks for putting this together! 👍 </p>",
      "rawMarkdown": "Cool! Thanks for putting this together! 👍",
      "votes": null
    },
    {
      "id": "652756",
      "postDate": "10/19/2019 11:12:07",
      "content": "<p>Very helpful. Please keep the thread updated.</p>\n\n<p>May you should replace all the links with proper titles, so that it becomes easy to follow the threads. Just my thought.</p>",
      "rawMarkdown": "Very helpful. Please keep the thread updated.\n\nMay you should replace all the links with proper titles, so that it becomes easy to follow the threads. Just my thought.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 652433,
      "author_name": "chmaxx",
      "author_url": "",
      "post_date": "10/18/2019 20:54:03",
      "content": "<p>Cool! Thanks for putting this together! 👍 </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 652756,
      "author_name": "arnabbiswas1",
      "author_url": "",
      "post_date": "10/19/2019 11:12:07",
      "content": "<p>Very helpful. Please keep the thread updated.</p>\n\n<p>May you should replace all the links with proper titles, so that it becomes easy to follow the threads. Just my thought.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "651301": "**Task**\nGiven hourly data of a building, predict future hourly data of building\n\n**Evaluation Metric**\nmetric is calculated once using all of the readings from all meter types\nsource: https://www.kaggle.com/c/ashrae-energy-prediction/discussion/113086#latest-650958 \nHow to calculate: https://www.kaggle.com/marknagelberg/rmsle-function , https://www.kaggle.com/jpopham91/rmlse-vectorized, \nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/113064#latest-650950 \n\n**Domain Knowledge / Getting Started**\nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112858#latest-649854 \nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/113080#latest-651137 \n\n**Similar Competitions**\nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112855#latest-649925 \nhttp://drivendata.co/blog/power-laws-forecasting-winners/ \nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112958#latest-650382 \n\n**Food for Thought**\nTrain only spans 1 year (Test is 2 years; therefore, is Time Series analysis less useful than Regression? But hosts say: “Our shared goals are to uncover (and disseminate to the public) current ‘best practices’ for methods of time series predictions that can be useful in our domain.” https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112985#latest-651212 )\nWhat is a negative precip_depth_1_hr value?\nHow will you handle &gt;30% missing data for floor_count, year_built, cloud_coverage, and precip_depth_1_hr ?\nActual weather vs forecasted weather: https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112887#latest-650545 , https://www.kaggle.com/c/ashrae-energy-prediction/discussion/112884#latest-651115 , https://www.mdpi.com/2504-4990/1/3/56/htm \n\n**Feature Importance**\nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112871#latest-650996 \n\n**Feature Engineering**\nDays (schools are usually closed during Saturday/Sunday)\nMonths (schools are closed during certain months)\nHolidays (many organizations may close on holidays)\n\n**Model Choice**\nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/113007#latest-650802 \nhttps://www.kaggle.com/c/ashrae-energy-prediction/discussion/112856#latest-650709 \n\n**Dealing with RAM usage**\nhttps://www.kaggle.com/hamditarek/reducing-memory-size-for-great-energy-predictor \nhttps://www.kaggle.com/yuliagm/how-to-work-with-big-datasets-on-16g-ram-dask",
    "652433": "Cool! Thanks for putting this together! 👍",
    "652756": "Very helpful. Please keep the thread updated.\n\nMay you should replace all the links with proper titles, so that it becomes easy to follow the threads. Just my thought."
  },
  "source": "meta"
}