In pandas 0.8.0, the time series plotting will be much improved (see attached example plot) What I'd love to see is an option to switch the reference time value off for aggregated time series.
Examples: - I have an average of wind speeds created from 5 years of data --> plotting a year would not make sense - daily maximum temperatures recoded during a single month --> depending on the case, showing the reference month or not could be desirable.
Comment From: timmie
The implementation shall aim at passing a string for formatting of label to the function:
ts.plot(dt_str='%D.%M')
Comment From: timmie
Maybe the subject could better generalised into:
allow modification of datetime formatting in output to - plots - files (would also allow to create a convenince function for solving #1337)
Comment From: changhiskhan
In mpl each data type is mapped to a single converter, so I think the way to do this is to give the custom format string to the x-axis and then have the converter check whether a custom format exists when getting the tick labels. However, this will mean that with custom format strings you'll lose the dynamic labeling that switches based on the axis scales.
Comment From: jorisvandenbossche
Closing this as there is no clear enhancement for pandas, and you can always change the default Formatter with matplotlib.