Plotting irregular timeseries with different indexes: last plotted timeseries dictates xaxis range

[lesteve]

In the following I took the extreme case where there is no overlap between the two indexes for illustration purposes. When there is some overlap there is still the same issue and the last plotted timeseries is dictating the xaxis range.

Here is the code, tested on pandas 0.13.1:

import numpy as np
import pandas as pd
import pandas.util.testing
import matplotlib.pyplot as plt

np.random.seed(0)

ts = pd.util.testing.makeTimeSeries()[:20]
tsIrregular = ts.drop(ts.index[[3,7,11,16]])

def myPlot(ts):
    fig, axes = plt.subplots(1,2)
    ax = axes[0]
    ax.set_title('left series first')
    ts[:5].plot(ax=ax, label='left series')
    ts[6:].plot(style='or', ax=ax, label='right series')
    ax.legend(loc='best')

    ax = axes[1]
    ax.set_title('right series first')
    ts[6:].plot(style='or', ax=ax, label='right series')
    ts[:5].plot(ax=ax, label='left series')
    ax.legend(loc='best')

    return axes

myPlot(ts)
myPlot(tsIrregular)

What you get for the regular timeseries case is what you expect, i.e. the order you plot timeseries doesn't matter and the xaxis range is the union of the two timeseries indexes:

but what you get for the irregular timeseries case is rather suprising:

There are easy work-arounds, like aligning the timeseries before plotting, but it feels like irregular timeseries shouldn't be second-class citizens compared to regular ones.

[jreback]

stale

[lesteve]

FWIW the behaviour is still the same in 0.13.1 (I slightly updated the code in the example to this effect)

I completely understand that it's not at the top of anyone's priority list though.

[jreback]

you are welcome to submit a pr

[rosnfeld]

Just wanted to note that I am working on this. A 1-line change can clear up the example presented above, all tests pass, and I think it's "correct", but unfortunately it then exposes a deeper version of the regular/irregular timeseries mismatch issue I noted in #6608. I'm tempted to try and fix both at once.