{
  "id": 116028,
  "title": "Missing last few hours of weather",
  "url": "/competitions/ashrae-energy-prediction/discussion/116028",
  "author_name": "",
  "post_date": "2019-11-06T17:08:10.346891900Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Since the weather is in GMT and some of the Sites are as far to the West as California, time conversion leads to missing data for the few last hours of 2018:\nLatest <code>weather_test</code>timestamp is <code>2018-12-31 23:00:00</code>, which becomes 15:00 in California (Site 4).\nLatest timestamp in <code>train</code> dataset is also <code>2018-12-31 23:00:00</code> but this is 8 hours later than the latest timestamp in <code>weather_test</code>.</p>\n\n<p>I see 2 obvious fixes: \n(1) looking up actual weather online\n(2) copying weather from 12/31/18 to 1/1/19/\nPS:  in case of <code>weather_train</code> dataset it is possible just to use weather data from the beginning of <code>weather_test</code>dataset.</p>\n\n<p>How do you deal with this? </p>",
  "messages": [
    {
      "id": "666965",
      "postDate": "11/06/2019 17:08:10",
      "content": "<p>Since the weather is in GMT and some of the Sites are as far to the West as California, time conversion leads to missing data for the few last hours of 2018:\nLatest <code>weather_test</code>timestamp is <code>2018-12-31 23:00:00</code>, which becomes 15:00 in California (Site 4).\nLatest timestamp in <code>train</code> dataset is also <code>2018-12-31 23:00:00</code> but this is 8 hours later than the latest timestamp in <code>weather_test</code>.</p>\n\n<p>I see 2 obvious fixes: \n(1) looking up actual weather online\n(2) copying weather from 12/31/18 to 1/1/19/\nPS:  in case of <code>weather_train</code> dataset it is possible just to use weather data from the beginning of <code>weather_test</code>dataset.</p>\n\n<p>How do you deal with this? </p>",
      "rawMarkdown": "Since the weather is in GMT and some of the Sites are as far to the West as California, time conversion leads to missing data for the few last hours of 2018:\nLatest `weather_test `timestamp is `2018-12-31 23:00:00`, which becomes 15:00 in California (Site 4).\nLatest timestamp in `train` dataset is also `2018-12-31 23:00:00` but this is 8 hours later than the latest timestamp in `weather_test`.\n\nI see 2 obvious fixes: \n(1) looking up actual weather online\n(2) copying weather from 12/31/18 to 1/1/19/\nPS:  in case of `weather_train` dataset it is possible just to use weather data from the beginning of `weather_test `dataset.\n\nHow do you deal with this?",
      "votes": null
    },
    {
      "id": "668050",
      "postDate": "11/07/2019 23:09:57",
      "content": "<p>You might have noticed that in my notebook <a href=\"https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans\">https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans</a> some of the data goes to 1/1/2019. I started doing this when I realised the same. All <a href=\"https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans#666525\">new data</a> I collect will have this 👍 </p>",
      "rawMarkdown": "You might have noticed that in my notebook https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans some of the data goes to 1/1/2019. I started doing this when I realised the same. All [new data](https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans#666525) I collect will have this 👍",
      "votes": null
    },
    {
      "id": "668316",
      "postDate": "11/08/2019 09:07:29",
      "content": "<p>thanks! this should fix it.</p>",
      "rawMarkdown": "thanks! this should fix it.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 668050,
      "author_name": "datadugong",
      "author_url": "",
      "post_date": "11/07/2019 23:09:57",
      "content": "<p>You might have noticed that in my notebook <a href=\"https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans\">https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans</a> some of the data goes to 1/1/2019. I started doing this when I realised the same. All <a href=\"https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans#666525\">new data</a> I collect will have this 👍 </p>",
      "votes": null,
      "replies": [
        {
          "id": 668316,
          "author_name": "poedator",
          "author_url": "",
          "post_date": "11/08/2019 09:07:29",
          "content": "<p>thanks! this should fix it.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "666965": "Since the weather is in GMT and some of the Sites are as far to the West as California, time conversion leads to missing data for the few last hours of 2018:\nLatest `weather_test `timestamp is `2018-12-31 23:00:00`, which becomes 15:00 in California (Site 4).\nLatest timestamp in `train` dataset is also `2018-12-31 23:00:00` but this is 8 hours later than the latest timestamp in `weather_test`.\n\nI see 2 obvious fixes: \n(1) looking up actual weather online\n(2) copying weather from 12/31/18 to 1/1/19/\nPS:  in case of `weather_train` dataset it is possible just to use weather data from the beginning of `weather_test `dataset.\n\nHow do you deal with this?",
    "668050": "You might have noticed that in my notebook https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans some of the data goes to 1/1/2019. I started doing this when I realised the same. All [new data](https://www.kaggle.com/datadugong/locate-better-cities-by-weather-temp-fill-nans#666525) I collect will have this 👍",
    "668316": "thanks! this should fix it."
  },
  "source": "meta"
}