Describe the missing or desired feature
TimeSeries.max(axis=0) and TimeSeries.max(axis=0) return a single-step timeseries, with the TimeIndex corresponding to the first timestep of the original series. This is documented correctly but is not intuitive. This seems to be done for consistency with Timeseries.mean(axis=0), median(axis=0), etc. There also does not seem to be a TimeSeries equivalent of Series.idxmax or np.ndarray.argmax.
Describe proposed solution
Chage max() and min() to return a single-step timeseries with the index of the returned value. This would be a breaking change obviously, though I cannot immediately imagine code depending on the previous behavior. Alternatively, add a keyword argument to enable this behavior, defaulting to current behavior.
Describe potential alternatives
Or alternatively, add Timeseries.idxmax/min/etc.
Describe the missing or desired feature
TimeSeries.max(axis=0)andTimeSeries.max(axis=0)return a single-step timeseries, with the TimeIndex corresponding to the first timestep of the original series. This is documented correctly but is not intuitive. This seems to be done for consistency withTimeseries.mean(axis=0),median(axis=0), etc. There also does not seem to be aTimeSeriesequivalent ofSeries.idxmaxornp.ndarray.argmax.Describe proposed solution
Chage
max()andmin()to return a single-step timeseries with the index of the returned value. This would be a breaking change obviously, though I cannot immediately imagine code depending on the previous behavior. Alternatively, add a keyword argument to enable this behavior, defaulting to current behavior.Describe potential alternatives
Or alternatively, add
Timeseries.idxmax/min/etc.