Fixes issue where a different datetime format is being used alongside clustering.
Oddly, if you use the exact=False flag in pandas.to_datetime it will manage 01/01/2050 00:00:00 with the format %Y-%m-%d %H:%M but not01/01/2050. That is, the "inexact" nature of the timestamp string is that it is longer than the dateformat, not shorter.
Hence, I tack on 00:00:00 to the clustering dates before converting them to datetime.
Summary of changes in this pull request:
If not using ISO standard dateformat, allow inexact matches.
If using time clustering, expect dateformat of the index and values to be in the model dateformat. I think this is the correct approach as the format is likely to be non-ISO because the CSV was opened in Excel, at which point any timestamp strings will be converted to the system format.
Fixes issue where a different datetime format is being used alongside clustering.
Oddly, if you use the
exact=False
flag in pandas.to_datetime it will manage01/01/2050 00:00:00
with the format%Y-%m-%d %H:%M
but not01/01/2050
. That is, the "inexact" nature of the timestamp string is that it is longer than the dateformat, not shorter.Hence, I tack on
00:00:00
to the clustering dates before converting them to datetime.Summary of changes in this pull request:
Reviewer checklist: