Grad-Cam CNN neural network method added

boxsers/__init__.py

@@ -1,5 +1,5 @@
 from boxsers.machine_learning import SpectroGmixture, SpectroKmeans, SpectroRF, SpectroSVM, \
-     SpectroLDA, SpectroPCA, SpectroCNN, cf_matrix, clf_report
+     SpectroLDA, SpectroPCA, SpectroCNN, validation_metrics
 from boxsers.data_augmentation import aug_mixup, aug_xshift, aug_noise, aug_multiplier, aug_linslope, aug_offset
 from boxsers.preprocessing import als_baseline_cor, savgol_smoothing, spectral_cut, spline_interpolation, \
     spectral_normalization, cosmic_filter

boxsers/machine_learning/classification.py

@@ -14,8 +14,7 @@
 import time
 import seaborn as sns
 import matplotlib.pyplot as plt
-from boxsers._boxsers_utils import _lightdark_switch
-from sklearn.inspection import permutation_importance
+import warnings
 
 
 class _MachineLearningClassifier:
@@ -134,6 +133,9 @@ def get_classif_report(self, x_test, y_test, digits=4, class_names=None, save_pa
         Returns:
             Scikit Learn classification report
         """
+        warnings.warn('\'get_classif_report\' instance method is no longer supported, use instead'
+                      ' the standalone \'clf_report\' method found in boxsers.validation_metrics module.',
+                      DeprecationWarning)
         # Converts binary labels to integer labels. Does nothing if they are already integer labels.
         if y_test.ndim == 2 and y_test.shape[1] > 1:
             y_test = np.argmax(y_test, axis=1)
@@ -199,6 +201,9 @@ def get_conf_matrix(self, x_test, y_test, normalize='true', class_names=None, ti
         Return:
             Scikit Learn confusion matrix
         """
+        warnings.warn('\'get_conf_matrix\' instance method is no longer supported, use instead'
+                      ' the standalone \'cf_matrix\' method found in boxsers.validation_metrics module.',
+                      DeprecationWarning)
         # The model returns the predicted classes
         y_pred = self.model.predict(x_test)
 
@@ -262,113 +267,6 @@ def __init__(self, n_trees=250, rdm_ste=None):
         # inherits the methods and arguments of the parent class _MachineLearningClassifier
         super().__init__(rf_model)
 
-    def plot_feat_importance(self, wn, sp, n_repeats=10, rdm_ste=None, title=None,
-                             xlabel='Raman Shift (cm$^{-1}$)', ylabel='Features importance',
-                             eline_width=1.5, marker_style='o', color=None, darktheme=False,
-                             grid=True, fontsize=10, fig_width=6.08, fig_height=3.8, save_path=None):
-        """ Plot feature importance based on permutation importance
-
-        Inspired by : L. Breiman, “Random Forests”, Machine Learning, 45(1), 5-32, 2001.
-
-        Parameters:
-            wn : array or list
-                X-axis(wavenumber, wavelenght, Raman shift, etc.), array shape = (n_pixels, ).
-
-            sp : array
-                Input Spectrum(s), array shape = (n_spectra, n_pixels) for multiple spectra and (n_pixels,)
-                for a single spectrum.
-
-            n_repeats : int, default=10
-                Number of times to permute a feature.
-
-            rdm_ste : integer, default=None
-                Random seed of the split. Using the same seed results in the same subsets
-                each time.
-
-            title : string, default=None
-                Font size(pts) used for the different elements of the graph. The title's font
-                 is two points larger than "fonctsize".
-
-            xlabel : string, default='Raman Shift (cm$^{-1}$)'
-                X-axis title. If None, there is no title displayed.
-
-            ylabel : string, default='Intensity (a.u.)'
-                Y-axis title. If None, there is no title displayed.
-
-            eline_width : positive float, default= 1.5
-                Plot errorbar line width(s).
-
-            marker_style : string, default='o'
-                Marker style.
-
-            color : string, default=None
-                Plot line color(s).
-
-            darktheme : boolean, default=False
-                If True, returns a plot with a dark background.
-
-            grid : boolean, default=False
-                If True, a grid is displayed.
-
-            fontsize : positive float, default=10
-                Font size(pts) used for the different elements of the graph.
-
-            fig_width : positive float or int, default=6.08
-                Figure width in inches.
-
-            fig_height : positive float or int, default=3.8
-                Figure height in inches.
-
-            save_path : string, default=None
-                Path where the figure is saved. If None, saving does not occur.
-                Recommended format : .png, .pdf
-        """
-        # update theme related parameters
-        frame_color, bg_color, alpha_value = _lightdark_switch(darktheme)
-
-        perm_imp = permutation_importance(self.model, wn, sp, n_repeats=n_repeats, random_state=rdm_ste)
-        imp_mean = perm_imp.importances_mean
-        imp_std = perm_imp.importances_std
-
-        # creates a figure object
-        fig = plt.figure(figsize=(fig_width, fig_height))
-        # add an axes object
-        ax = fig.add_subplot(1, 1, 1)  # nrows, ncols, index
-
-        ax.errorbar(range(50), imp_mean, yerr=imp_std, fmt='none', ecolor=frame_color, elinewidth=eline_width,
-                    capsize=5, alpha=0.85)
-        ax.plot(range(50), imp_mean, marker_style, color=color)
-
-        # titles settings
-        ax.set_title(title, fontsize=fontsize + 1.2,
-                     color=frame_color)  # sets the plot title, 1.2 points larger font size
-        ax.set_xlabel(xlabel, fontsize=fontsize, color=frame_color)  # sets the X-axis label
-        ax.set_ylabel(ylabel, fontsize=fontsize, color=frame_color)  # sets the Y-axis label
-
-        # tick settings
-        ax.minorticks_on()
-        ax.tick_params(axis='both', which='major',
-                       labelsize=fontsize - 2,  # 2.0 points smaller font size
-                       color=frame_color)
-        ax.tick_params(axis='both', which='minor', color=frame_color)
-        ax.tick_params(axis='x', colors=frame_color)  # setting up X-axis values color
-        ax.tick_params(axis='y', colors=frame_color)  # setting up Y-axis values color
-        for spine in ['top', 'bottom', 'left', 'right']:
-            ax.spines[spine].set_color(frame_color)  # setting up spines color
-        if grid is True:
-            # adds a grid
-            ax.grid(alpha=alpha_value)
-
-        # set figure and axes facecolor
-        fig.set_facecolor(bg_color)
-        ax.set_facecolor(bg_color)
-        # adjusts subplot params so that the subplot(s) fits in to the figure area
-        fig.tight_layout()
-        # save figure
-        if save_path is not None:
-            plt.savefig(save_path, dpi=300, bbox_inches='tight')
-        plt.show()
-
 
 class SpectroSVM(_MachineLearningClassifier):
     """ Support Vector Machine classification model.

boxsers/machine_learning/dimension_reduction.py

@@ -125,8 +125,8 @@ def pair_plot(self, sp, lab, n_components=1,  class_names=None, title=None,
 
         # title settings
         ax.set_title(title, fontsize=fontsize + 1.2, color=frame_color)  # 1.2 points larger font size
-        ax.set_xlabel(fontsize=fontsize, color=frame_color)  # sets the x-axis title
-        ax.set_ylabel(fontsize=fontsize, color=frame_color)  # sets the y-axis title
+        # ax.set_xlabel(fontsize=fontsize, color=frame_color)  # sets the x-axis title
+        # ax.set_ylabel(fontsize=fontsize, color=frame_color)  # sets the y-axis title
 
         # tick settings
         ax.minorticks_on()