ds = xr.open_dataset(fname)
features = [var for var in list(ds.data_vars) if 'matched' not in var
] + ['Run Date']
readable_feature_names = {
feature: to_readable_names([feature])[0]
for feature in features
}
feature_colors = {
feature: to_readable_names([feature])[1]
for feature in features
}
xticks = np.arange(0, 1.2, 0.2)
method_bool = True if method == 'multipass' else False
fig = myInterpreter.plot_importance(
multipass=method_bool,
display_feature_names=readable_feature_names,
feature_colors=feature_colors,
num_vars_to_plot=num_vars_to_plot,
xticks=xticks,
metric=metric.upper(),
metrics_used=ylabels,
ylabels=ylabels,
xlabels=['NAUPDC'],
n_columns=len(ml_models))
fname = f'{method}_permutation_importance_{time}_{mode}_{drop_opt}_compare.png'
base_plot.save_figure(fig=fig, fname=fname)
feature: to_readable_names([feature])[1]
for feature in features
}
adict = readable_feature_names
display_feature_names1 = {
f'{f[0]}__{f[1]}': f'{adict[f[0]]} & {adict[f[1]]}'
for f in list(itertools.combinations(features, r=2))
}
display_feature_names2 = {
f'{f[1]}__{f[0]}': f'{adict[f[0]]} & {adict[f[1]]}'
for f in list(itertools.combinations(features, r=2))
}
display_feature_names = {**display_feature_names1, **display_feature_names2}
fig = myInterpreter.plot_importance(
data=results,
method='perm_based_interactions',
model_output='probability',
display_feature_names=display_feature_names,
num_vars_to_plot=num_vars_to_plot,
ylabels=ylabels,
figsize=figsize,
plot_correlated_features=True,
fontsize=5)
fname = f'perm_based_interact_rankings_{time}_{drop_opt}.png'
base_plot.save_figure(fig=fig, fname=fname)
for target in targets:
fnames = [
get_fnames(model_name, target, time, drop_opt)
for model_name in ml_models
]
results.append(myInterpreter.load_results(fnames))
readable_feature_names = {
feature: to_readable_names([feature])[0]
for feature in feature_names
}
feature_colors = {
feature: to_readable_names([feature])[1]
for feature in feature_names
}
fig = myInterpreter.plot_importance(
data=results,
method='ale_variance',
model_output='probability',
display_feature_names=readable_feature_names,
feature_colors=feature_colors,
num_vars_to_plot=num_vars_to_plot,
ylabels=ylabels,
figsize=figsize,
plot_correlated_features=True,
)
fname = f'ale_importance_{time}_{drop_opt}.png'
base_plot.save_figure(fig=fig, fname=fname)
