tsplot crashes when there is only one datapoint

when data has only one datapoint (for a given unit and condition), tsplot crashes with an out of bounds exception.

seaborn/timeseries.py in tsplot(data, time, unit, condition, value, err_style, ci, interpolate, color, estimator, n_boot, err_palette, err_kws, legend, ax, **kwargs)
    328     # Pad the sides of the plot only when not interpolating
    329     ax.set_xlim(x.min(), x.max())
--> 330     x_diff = x[1] - x[0]
    331     if not interpolate:
    332         ax.set_xlim(x.min() - x_diff, x.max() + x_diff)

IndexError: index 1 is out of bounds for axis 0 with size 1

[mwaskom]

Closing with tsplot deprecation.

[BlackHC]

https://cloud.sagemath.com/projects/26d7dccf-e180-450e-b348-05e3242d8779/files/2016-12-14-171122.ipynb

import pandas as pd
import seaborn

# This crashes.
data = [
    {'unit': 1, 'condition': '2015-12-02', 'tick': 1, 'value': 2},  
    {'unit': 1, 'condition': '2015-12-02', 'tick': 2, 'value': 3},
    {'unit': 1, 'condition': '2015-12-01', 'tick': 1, 'value': 1}, 
]
data_frame = pd.DataFrame(data)
seaborn.tsplot(data_frame, time="tick", unit="unit", condition="condition", value="value", err_style="unit_traces")

Creating the example, I noticed that it only breaks if the single data point trace comes last because x is a loop variable, so the padding only takes the last trace into account… which means that the padding is probably off when you have traces that vary wildly in their span in general.

Thanks, Andreas