Updated_Peak_Ratio

[daniel-caichac-DHI]

Hi

• We updated the definition of PR, instead of just the mean of peaks of model / mean of peaks of obs , it is now calculated as the mean of peak_mod/peak_meas (it, the mean of the individual ratios, it is not a linear relationship so values change slightly with previous results.
• Also, before it was calculated with all detected peaks, now it is only calculated with the joint-peaks, ie, concurrent peaks within +- 0.5*inter_event_time (default = 36h, hence PR now only depends on joint peaks as long as they ocurr within +- 18h).
• This could mean that the Peak Ratio could be Nan if no joint-peaks are found, in which case a warning is given and None is returned.
• Updated documentation and tests (had to upload a new test file)

[daniel-caichac-DHI]

@ecomodeller Could you help me with the tests? in my laptop pytest is successful, but on the web there are some extra checks even before pytest that crash, and I do not know how to fix.

[ecomodeller]

@daniel-caichac-DHI take a look again (and don't return None)

[ecomodeller]

Don't return None.

Either raise an Error, or maybe return np.nan.

[daniel-caichac-DHI]

There bwas a conflict.
I resolved it but keeps failing with these new thing of chartboost , it does not even start the pytest
:(

[daniel-caichac-DHI]

Updated the tests (forgot before)

[daniel-caichac-DHI]

All good now apparently :)

[daniel-caichac-DHI]

Last commit was just to format the test as you did last time