Here is one sample:
102406 “weekly”
We have test data up to [1] 2016-02-04 23:00:00, and submission timestamps:
[2] 2016-02-11 00:00:00
[3] 2016-02-18 00:00:00
between [2] and [3], there are 7 days, i.e., 2/11,2/12,2/13,2/14,2/15,2/16, and 2/17
However, between [1] and [2], there are only 6 days, i.e., 2/5,2/6,2/7,2/8,2/9, and 2/10.
Is this intentional or by mistake?
All weekly prediction times have this problem, and when we strictly follow the definition of consumption (up to that timestamp, from the previous timestamp), we need to predict consumption of 6 days for the first week, and 7 days for the second week.
Please clarify this glitch.
Thanks.