things has been changed

TASK_4/lime_irene.py

@@ -85,7 +85,7 @@ def get_lime_importance_from_explanation(lime_explanation, prediction):
 # - discretize_continuous = True (continuous features are discretized into quartiles)
 # - discretizer = 'quartile'
 # - sample_around_instance = False (sample from a normal centered on the mean of the feature data)
-# 
+#
 # Also to explain instances we will use deault parameters, i.e.:
 # - num_samples = 5000, number of samples to generate
 # - distance_metric = 'euclidean', distance metric to use for weights
@@ -273,12 +273,6 @@ def get_lime_importance_from_explanation(lime_explanation, prediction):
 # %%
 clf_names = [Classifiers.DT.value, Classifiers.RF.value, Classifiers.SVM.value, Classifiers.XGB.value, Classifiers.NN.value]
 
-# %%
-classifiers[Classifiers.NN.value].predict_proba(indicators_test_db_df[:10])
-
-# %%
-classifiers[Classifiers.RF.value].predict_proba(indicators_test_rb_df[:10])
-
 # %%
 positions_to_explain = selected_records_to_explain_df['positions'].to_list()
 instance_names_to_explain = selected_records_to_explain_df['instance names'].to_list()
@@ -300,7 +294,7 @@ def get_lime_importance_from_explanation(lime_explanation, prediction):
     clf_metrics = {}
     for i in range(instances.shape[0]):
         prediction = classifier.predict(instances[i].reshape(1,-1))[0]
-        if clf_name == Classifiers.SVM.value:
+        if clf_name == Classifiers.SVM.value or clf_name == Classifiers.NN.value:
             prob = classifier.predict_proba(instances[i].reshape(1,-1))[0]
             if prob[1] >= 0.5:
                 prediction = 1
@@ -331,12 +325,6 @@ def get_lime_importance_from_explanation(lime_explanation, prediction):
 
 metrics_selected_records_df
 
-# %%
-explanation.local_exp[0]
-# feature_importances = np.zeros(len(pred_explanation))
-# for tuple in pred_explanation:
-#     feature_importances[tuple[0]] = tuple[1]
-
 # %%
 random_records_to_explain_df = pd.read_csv('../data/explanation_results/random_records_to_explain.csv', index_col=0)
 positions_to_explain = random_records_to_explain_df['positions'].to_list()
@@ -357,7 +345,7 @@ def get_lime_importance_from_explanation(lime_explanation, prediction):
     faithfulness = []
     for i in range(instances.shape[0]):
         prediction = classifier.predict(instances[i].reshape(1,-1))[0]
-        if clf_name == Classifiers.SVM.value:
+        if clf_name == Classifiers.SVM.value or clf_name == Classifiers.NN.value:
             prob = classifier.predict_proba(instances[i].reshape(1,-1))[0]
             if prob[1] >= 0.5:
                 prediction = 1