Plot comparison test failing

[wsavran]

On line 1486 in csep/utils/plots.py the plot_comparison_test is failing due to an incompatible change in matplotlib. The fix currently exists in plot_poisson_consistency_test and should be duplicated in plot_comparison_test.

[pabloitu]

can't seem to reproduce this bug :C. I think the comparison plot was not affected by the matplotlib change. The change forced errorbars to be defined only by positive values (e.g. plt.errorbar(origin, xerr=[1, 2]) giving a (-1, +2) error margin), meaning the origin point now must be inside the error range. That was the problem with consistency tests (with obs outside sim range), but t-tests was already having the origin inside by default

created a unit test for comparison tests in, #222 that randomizes the values to see if at some point it breaks. so far it works alright

[pabloitu]

should we close this?

[wsavran]

UPDATE: Open pull request to address this. @pabloitu @fabiolsilva let me know what you think.

im reopening this because I'm seeing this is still an issue.

see example script to reproduce

import csep
from csep.core import poisson_evaluations as poisson
from csep.utils import datasets, time_utils, plots

# Needed to show plots from the terminal
import matplotlib.pyplot as plt

start_date = time_utils.strptime_to_utc_datetime('2006-11-12 00:00:00.0')
end_date = time_utils.strptime_to_utc_datetime('2011-11-12 00:00:00.0')

forecast = csep.load_gridded_forecast(datasets.helmstetter_aftershock_fname,
                                      start_date=start_date,
                                      end_date=end_date,
                                      name='helmstetter_aftershock')

print("Querying comcat catalog")
catalog = csep.query_comcat(forecast.start_time, forecast.end_time, min_magnitude=forecast.min_magnitude)
print(catalog)

catalog = catalog.filter_spatial(forecast.region)
print(catalog)

t_test = poisson.paired_t_test(forecast, forecast, catalog)

ax = plots.plot_comparison_test([t_test], plot_args={})

This script is failing with the following error.

/Users/wsavran/Working/github/pycsep/csep/utils/plots.py:1761: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.
  ax.set_xticklabels([res.sim_name[0] for res in results_t],
Traceback (most recent call last):
  File "/Users/wsavran/Desktop/test.py", line 25, in <module>
    ax = plots.plot_comparison_test([t_test], plot_args={})
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/wsavran/Working/github/pycsep/csep/utils/plots.py", line 1773, in plot_comparison_test
    width=(xTickPos[1] - xTickPos[0]), color=['gray', 'w'], alpha=0.2)
           ~~~~~~~~^^^
IndexError: index 1 is out of bounds for axis 0 with size 1

some notes:

1. this issue only happens when len(results_t) == 1
2. @pabloitu you're right this wasn't related to matplotlib changes, but that we're shading the regions and not checking if there is more than 2 forecasts when doing this.