refactor: t-test EvaluationResult() now includes value of alpha, so i…

…t is not necessary to set it again when plotting. plot_comparison_test() includes a default legend that explains the symbology therein.

csep/core/poisson_evaluations.py

@@ -47,7 +47,7 @@ def paired_t_test(forecast, benchmark_forecast, observed_catalog,
     result.name = 'Paired T-Test'
     result.test_distribution = (out['ig_lower'], out['ig_upper'])
     result.observed_statistic = out['information_gain']
-    result.quantile = (out['t_statistic'], out['t_critical'])
+    result.quantile = (out['t_statistic'], out['t_critical'], alpha)
     result.sim_name = (forecast.name, benchmark_forecast.name)
     result.obs_name = observed_catalog.name
     result.status = 'normal'

csep/utils/plots.py

@@ -773,7 +773,8 @@ def plot_comparison_test(
     if plot_args["legend"]:
         # Add custom legend to explain results
         legend_elements = [
-            Line2D([0], [0], color="red", lw=2, label="T-test rejected"),
+            Line2D([0], [0], color="red", lw=2,
+                   label=f"T-test rejected ($\\alpha = {results_t[0].quantile[-1]}$)"),
             Line2D([0], [0], color="green", lw=2, label="T-test non-rejected"),
             Line2D([0], [0], color="gray", lw=2, label="T-test indistinguishable"),
             Line2D(

tests/test_plots.py

@@ -662,7 +662,7 @@ def rand(limit=10, offset=0.0):
                     t_result.observed_statistic - rand(5),
                     t_result.observed_statistic + rand(5),
                 ]
-
+                t_result.quantile = (None, None, 0.05)
                 if numpy.random.random() < 0.05:  # sim possible infinite values
                     t_result.observed_statistic = -numpy.inf
                 t_tests.append(t_result)