[MAX STABLE] add starting point for non stationary full estimator. fix display...

[MAX STABLE] add starting point for non stationary full estimator. fix display in non stationary trends

experiment/meteo_france_SCM_study/visualization/study_visualization/non_stationary_trends.py

@@ -81,16 +81,12 @@ class AbstractNonStationaryTrendTest(object):
         mu1_trends = [self.get_mu1(starting_point=year) for year in years]
         ax2 = ax.twinx()
         color_mu1 = 'c'
-        ax.plot(years, mu1_trends, color_mu1 + 'o-')
+        ax2.plot(years, mu1_trends, color_mu1 + 'o-')
         ax2.set_ylabel('mu1 parameter', color=color_mu1)
 
         ax.set_xlabel('starting year for the linear trend of mu1')
         align_yaxis_on_zero(ax, ax2)
         title = self.display_name
-        first = mu1_trends[0]
-        last = mu1_trends[-1]
-        title += '\n mu1: first {}\nlast {}'.format(first, last)
-        title += '; equals? {}'.format(first == last)
         ax.set_title(title)
         ax.legend()
 

extreme_estimator/estimator/full_estimator/abstract_full_estimator.py

@@ -56,12 +56,21 @@ class FullEstimatorInASingleStepWithSmoothMargin(AbstractFullEstimator):
     def margin_function_start_fit(self):
         return self.linear_margin_model.margin_function_start_fit
 
+    @property
+    def df_coordinates_spat(self):
+        return self.dataset.coordinates.df_spatial_coordinates(self.train_split)
+
+    @property
+    def df_coordinates_temp(self):
+        df_coordinates_temp = self.dataset.coordinates.df_temporal_coordinates(self.train_split)
+        return self.linear_margin_model.add_starting_temporal_point(df_coordinates_temp)
+
     def _fit(self):
         # Estimate both the margin and the max-stable structure
         self._result_from_fit = self.max_stable_model.fitmaxstab(
             data_gev=self.dataset.maxima_gev_for_spatial_extremes_package(self.train_split),
-            df_coordinates_spat=self.dataset.coordinates.df_spatial_coordinates(self.train_split),
-            df_coordinates_temp=self.dataset.coordinates.df_temporal_coordinates(self.train_split),
+            df_coordinates_spat=self.df_coordinates_spat,
+            df_coordinates_temp=self.df_coordinates_temp,
             fit_marge=True,
             fit_marge_form_dict=self.linear_margin_model.margin_function_start_fit.form_dict,
             margin_start_dict=self.linear_margin_model.margin_function_start_fit.coef_dict

extreme_estimator/estimator/margin_estimator/abstract_margin_estimator.py

@@ -32,6 +32,7 @@ class LinearMarginEstimator(AbstractMarginEstimator):
         maxima_gev_specialized = self.dataset.maxima_gev_for_spatial_extremes_package(self.train_split)
         df_coordinates_spat = self.dataset.coordinates.df_spatial_coordinates(self.train_split)
         df_coordinates_temp = self.dataset.coordinates.df_temporal_coordinates(self.train_split)
+        df_coordinates_temp = self.margin_model.add_starting_temporal_point(df_coordinates_temp)
         self._result_from_fit = self.margin_model.fitmargin_from_maxima_gev(data=maxima_gev_specialized,
                                                                             df_coordinates_spat=df_coordinates_spat,
                                                                             df_coordinates_temp=df_coordinates_temp)

extreme_estimator/extreme_models/margin_model/parametric_margin_model.py

@@ -38,9 +38,6 @@ class ParametricMarginModel(AbstractMarginModel, ABC):
                                   df_coordinates_temp: pd.DataFrame) -> ResultFromFit:
         assert data.shape[1] == len(df_coordinates_spat)
 
-        # Modify df_coordinates_temp
-        df_coordinates_temp = self.add_starting_temporal_point(df_coordinates_temp)
-
         # Margin formula for fitspatgev
         fit_params = get_margin_formula(self.margin_function_start_fit.form_dict)
 

test/test_extreme_estimator/test_extreme_models/test_max_stable_temporal.py

@@ -99,14 +99,6 @@ class TestMaxStableTemporal(unittest.TestCase):
         # Create two different estimators
         estimator1 = self.fit_non_stationary_estimator(starting_point=3)
         estimator2 = self.fit_non_stationary_estimator(starting_point=20)
-
-        for starting_point in range(3, 20):
-            estimator = self.fit_non_stationary_estimator(starting_point=starting_point)
-            print(estimator.margin_function_fitted.starting_point)
-            print(estimator.margin_function_fitted.coef_dict)
-            print(estimator.margin_function_fitted.mu0)
-            print(estimator.margin_function_fitted.mu1_temporal_trend)
-
         mu1_estimator1 = estimator1.margin_function_fitted.mu1_temporal_trend
         mu1_estimator2 = estimator2.margin_function_fitted.mu1_temporal_trend
         self.assertNotEqual(mu1_estimator1, mu1_estimator2)