![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F255380%2F69af7656912cef230a07bf085a0a6564%2Fex1_daily.png?generation=1573386582938502&alt=media\")
There is a worrying pattern in the reported measurements for site_id 9 during the July 4th holidays. Most buildings have a single reading for each meter during the days of 1, 2, 3 and 4 of July that is normally at 1 am, and show 0 for all other hourly measurments during that day. It appears that for those days a daily sum has been reported instead of the hourly values. For example, for this specific building it is clear that there is something wrong with some days in July (colours correspond to June, July, August, sorry for the patchy graphs!):
\n\n
And if we summarize the daily sums, it is evident that what is given as hourly measurement is probable the daily sum:
\n\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F255380%2F6e8b44237c98138b3dc781e17724a040%2Fex1_sums.png?generation=1573386704724056&alt=media\")
Why this is done escapes me, but it certainly affects how we deal with these values.\nWould we assume that they report it this way in the next years?\nEven if they do, the 4th of July long weekend will not be there in the next two years, so it will not affect the same 4-5 days concerned.
", "messages": [ { "id": 669738, "postDate": "2019-11-10T12:15:53.510Z", "content": "There is a worrying pattern in the reported measurements for site_id 9 during the July 4th holidays. Most buildings have a single reading for each meter during the days of 1, 2, 3 and 4 of July that is normally at 1 am, and show 0 for all other hourly measurments during that day. It appears that for those days a daily sum has been reported instead of the hourly values. For example, for this specific building it is clear that there is something wrong with some days in July (colours correspond to June, July, August, sorry for the patchy graphs!):
\n\n
And if we summarize the daily sums, it is evident that what is given as hourly measurement is probable the daily sum:
\n\n
Why this is done escapes me, but it certainly affects how we deal with these values.\nWould we assume that they report it this way in the next years?\nEven if they do, the 4th of July long weekend will not be there in the next two years, so it will not affect the same 4-5 days concerned.
", "rawMarkdown": "There is a worrying pattern in the reported measurements for site_id 9 during the July 4th holidays. Most buildings have a single reading for each meter during the days of 1, 2, 3 and 4 of July that is normally at 1 am, and show 0 for all other hourly measurments during that day. It appears that for those days a daily sum has been reported instead of the hourly values. For example, for this specific building it is clear that there is something wrong with some days in July (colours correspond to June, July, August, sorry for the patchy graphs!):\n\n\n\n\nAnd if we summarize the daily sums, it is evident that what is given as hourly measurement is probable the daily sum:\n\n\n\nWhy this is done escapes me, but it certainly affects how we deal with these values.\nWould we assume that they report it this way in the next years?\nEven if they do, the 4th of July long weekend will not be there in the next two years, so it will not affect the same 4-5 days concerned. ", "votes": 14 } ], "comments": [], "raw_markdown_by_id": { "669738": "There is a worrying pattern in the reported measurements for site_id 9 during the July 4th holidays. Most buildings have a single reading for each meter during the days of 1, 2, 3 and 4 of July that is normally at 1 am, and show 0 for all other hourly measurments during that day. It appears that for those days a daily sum has been reported instead of the hourly values. For example, for this specific building it is clear that there is something wrong with some days in July (colours correspond to June, July, August, sorry for the patchy graphs!):\n\n\n\n\nAnd if we summarize the daily sums, it is evident that what is given as hourly measurement is probable the daily sum:\n\n\n\nWhy this is done escapes me, but it certainly affects how we deal with these values.\nWould we assume that they report it this way in the next years?\nEven if they do, the 4th of July long weekend will not be there in the next two years, so it will not affect the same 4-5 days concerned. " } }