fixed confusion plots

exp_xgb.ini → exp_emodb.ini

@@ -1,5 +1,5 @@
 [EXP]
-root = /home/fburkhardt/ResearchProjects/nkululeko
+root = /home/fburkhardt/ResearchProjects/nkululeko/
 store = ./store/
 name = xgb-exp
 fig_dir = ./images/
@@ -8,7 +8,7 @@ databases = ['emodb']
 emodb = /home/fburkhardt/audb/emodb/6/
 emodb.mapping = {'anger':'angry', 'happiness':'happy', 'sadness':'sad', 'neutral':'neutral'}
 emodb.split_strategy = speaker_split
-emodb.testsplit = 10
+emodb.testsplit = 40
 target = emotion
 labels = ['neutral', 'happy', 'sad', 'angry']
 [FEATS]

main.py → exp_emodb.py

@@ -38,4 +38,4 @@ def main(config_file):
 
 
 if __name__ == "__main__":
-    main('/home/fburkhardt/ResearchProjects/nkululeko/exp_xgb.ini') #sys.argv[1])
+    main('/home/fburkhardt/ResearchProjects/nkululeko/exp_emodb.ini') #sys.argv[1])

src/reporter.py

@@ -7,6 +7,8 @@
 import ast
 import numpy as np
 import glob_conf
+from sklearn.metrics import ConfusionMatrixDisplay
+from sklearn.metrics import confusion_matrix
 
 class Reporter:
 
@@ -20,12 +22,25 @@ def continuous_to_categorical(self):
         self.truths = np.digitize(self.truths, bins)-1
         self.preds = np.digitize(self.preds, bins)-1
 
+    def plot_confmatrix_mew(self, plot_name): 
+        fig_dir = self.util.get_path('fig_dir')
+        labels = ast.literal_eval(glob_conf.config['DATA']['labels'])
+        plt.figure()  # figsize=[5, 5]
+        cm = confusion_matrix(self.truths, self.preds,  normalize = 'true')
+        disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=labels).plot(cmap='Blues')
+        print(f'plotting conf matrix to {fig_dir+plot_name}')
+        plt.title('Confusion Matrix')
+        plt.savefig(fig_dir+plot_name)
+        plt.close()
+
+
     def plot_confmatrix(self, plot_name): 
         fig_dir = self.util.get_path('fig_dir')
         sns.set()  # get prettier plots
         labels = ast.literal_eval(glob_conf.config['DATA']['labels'])
-        plt.figure(figsize=[5, 5])
+        plt.figure()  # figsize=[5, 5]
         plt.title('Confusion Matrix')
+        plt.ylabel('UAR')
         audplot.confusion_matrix(self.truths, self.preds)
         # replace labels
         locs, _ = plt.xticks()
@@ -35,6 +50,12 @@ def plot_confmatrix(self, plot_name):
         print(f'plotting conf matrix to {fig_dir+plot_name}')
         plt.savefig(fig_dir+plot_name)
         plt.close()
+        print('truths')
+        print(self.truths.values)
+        print('preds')
+        print(self.preds)
+        print('labels')
+        print(labels)
 
     def uar(self):
         return recall_score(self.truths, self.preds, average='macro')