[contrasting] add plot mean, and plot relative change in the mean

projects/altitude_spatial_model/altitudes_fit/main_altitudes_studies.py

@@ -11,7 +11,10 @@ from projects.altitude_spatial_model.altitudes_fit.one_fold_analysis.altitudes_s
 def plot_altitudinal_fit(studies):
     visualizer = AltitudesStudiesVisualizerForNonStationaryModels(studies=studies,
                                                                   model_classes=ALTITUDINAL_MODELS,
-                                                                  massif_names=['Belledonne'])
+                                                                  massif_names=['Belledonne'],
+                                                                  show=True)
+    visualizer.plot_mean()
+    visualizer.plot_relative_change()
 
 
 def plot_time_series(studies):

projects/altitude_spatial_model/altitudes_fit/one_fold_analysis/altitudes_studies_visualizer_for_non_stationary_models.py

 from typing import List
+import matplotlib.pyplot as plt
+
+import numpy as np
 
 from extreme_data.meteo_france_data.scm_models_data.abstract_study import AbstractStudy
+from extreme_data.meteo_france_data.scm_models_data.visualization.main_study_visualizer import \
+    SCM_STUDY_CLASS_TO_ABBREVIATION
+from extreme_data.meteo_france_data.scm_models_data.visualization.plot_utils import plot_against_altitude
 from extreme_data.meteo_france_data.scm_models_data.visualization.study_visualizer import StudyVisualizer
 from extreme_fit.model.margin_model.polynomial_margin_model.spatio_temporal_polynomial_model import \
     AbstractSpatioTemporalPolynomialModel
@@ -24,4 +30,31 @@ class AltitudesStudiesVisualizerForNonStationaryModels(StudyVisualizer):
         for massif_name in self.massif_names:
             dataset = studies.spatio_temporal_dataset(massif_name=massif_name)
             old_fold_fit = OneFoldFit(dataset, model_classes)
-            self.massif_name_to_one_fold_fit[massif_name] = old_fold_fit
\ No newline at end of file
+            self.massif_name_to_one_fold_fit[massif_name] = old_fold_fit
+
+    def plot_mean(self):
+        self.plot_general('mean')
+
+    def plot_relative_change(self):
+        self.plot_general('relative_changes_in_the_mean')
+
+    def plot_general(self, method_name):
+        ax = plt.gca()
+        min_altitude, *_, max_altitude = self.studies.altitudes
+        altitudes = np.linspace(min_altitude, max_altitude, num=50)
+        for massif_id, massif_name in enumerate(self.massif_names):
+            old_fold_fit = self.massif_name_to_one_fold_fit[massif_name]
+            values = old_fold_fit.__getattribute__(method_name)(altitudes)
+            plot_against_altitude(altitudes, ax, massif_id, massif_name, values)
+        # Plot settings
+        ax.legend(prop={'size': 7}, ncol=3)
+        moment = ' '.join(method_name.split('_'))
+        plot_name = '{} annual maxima of {}'.format(moment, SCM_STUDY_CLASS_TO_ABBREVIATION[self.studies.study_class])
+        ax.set_ylabel('{} ({})'.format(plot_name, self.study.variable_unit), fontsize=15)
+        ax.set_xlabel('altitudes', fontsize=15)
+        # lim_down, lim_up = ax.get_ylim()
+        # lim_up += (lim_up - lim_down) / 3
+        # ax.set_ylim([lim_down, lim_up])
+        ax.tick_params(axis='both', which='major', labelsize=13)
+        self.studies.show_or_save_to_file(plot_name=plot_name, show=self.show)
+        ax.clear()

projects/altitude_spatial_model/altitudes_fit/one_fold_analysis/one_fold_fit.py

+import numpy as np
+from cached_property import cached_property
+
 from extreme_fit.estimator.margin_estimator.utils import fitted_linear_margin_estimator_short
 from extreme_fit.model.margin_model.utils import MarginFitMethod
 
@@ -18,3 +21,35 @@ class OneFoldFit(object):
         # Some display
         for estimator in self.model_class_to_estimator.values():
             print(estimator.result_from_model_fit.aic)
+
+    @cached_property
+    def best_estimator(self):
+        sorted_estimators = sorted([estimator for estimator in self.model_class_to_estimator.values()],
+                                   key=lambda e: e.result_from_model_fit.aic)
+        estimator_that_minimizes_aic = sorted_estimators[0]
+        return estimator_that_minimizes_aic
+
+    @property
+    def best_function_from_fit(self):
+        return self.best_estimator.function_from_fit
+
+    def mean(self, altitudes, year=2019):
+        return [self.get_mean(altitude, year) for altitude in altitudes]
+
+    def get_mean(self, altitude, year):
+        coordinate = np.array([altitude, year])
+        gev_params = self.best_function_from_fit.get_params(coordinate, is_transformed=False)
+        return gev_params.mean
+
+    def relative_changes_in_the_mean(self, altitudes, year=2019, nb_years=50):
+        relative_changes = []
+        for altitude in altitudes:
+            mean_after = self.get_mean(altitude, year)
+            mean_before = self.get_mean(altitude, year - nb_years)
+            relative_change = 100 * (mean_after - mean_before) / mean_before
+            relative_changes.append(relative_change)
+        return relative_changes
+
+
+
+