diff --git a/Code/Knn.py b/Code/Knn.py
index a7c8112..f5bab4e 100644
--- a/Code/Knn.py
+++ b/Code/Knn.py
@@ -111,6 +111,7 @@ class Knn: #nom de la class à changer
         """
         n,m = self.train.shape
         nn=np.empty((self.k,2))
+        nn[:,0]=[i for i in range(self.k)]
         nn[:,1]=np.apply_along_axis(minkowski_distance,1,self.train[0:self.k],exemple,self.minkowski_p)
         for i in range(self.k,n):
             dist = minkowski_distance(self.train[i],exemple,self.minkowski_p)
@@ -150,6 +151,32 @@ class Knn: #nom de la class à changer
         
         """
         
+        n,m = test.shape
+        n_ex,m_ex = self.train.shape
+        nn=np.empty((n,self.k,2))
+        # Boucle sur chaque ligne du jeu de test
+        for x in range(n):
+            nn[x,:,0]=[i for i in range(self.k)]
+            nn[x,:,1]=np.apply_along_axis(minkowski_distance,1,self.train[0:self.k],test[x],self.minkowski_p)
+            for i in range(self.k,n_ex):
+                dist = minkowski_distance(self.train[i],test[x],self.minkowski_p)
+                nn_dist=nn[x,:,1]
+                distdiff = nn_dist-dist
+                max_distdiff=max(distdiff)
+                if(max_distdiff>0):
+                    pos_changement = np.argwhere(nn_dist==max(nn_dist))[0]
+                    nn[x,pos_changement,0]=i
+                    nn[x,pos_changement,1]=max_distdiff
+        nn_labels = self.train_labels[nn[:,:,0].astype(np.int)]
+        nn_mode_label = np.apply_along_axis(mode,1,nn_labels)
+        # on construit la matrice de confusion
+        cm = self.confusion_matrix(test_labels,nn_mode_label)
+        accuracy, precision, recall = self.prediction_metrics(cm,test_labels,nn_mode_label)
+        self.print_prediction_metrics(cm,accuracy,precision,recall)
+
+        return cm,accuracy,precision,recall
+        
+        
     def confusion_matrix(self,obs_labels,pred_labels):
         """
         Retourne la matrice de confusion
@@ -157,11 +184,9 @@ class Knn: #nom de la class à changer
         Retourne une matrice NumPy
         """
         unique_obs_labels=np.unique(obs_labels)
-        unique_pred_labels=np.unique(pred_labels)
         nb_unique_obs_labels=(unique_obs_labels.shape)[0]
-        nb_unique_pred_labels=(unique_pred_labels.shape)[0]
         
-        confusion_matrix = np.zeros((nb_unique_obs_labels,nb_unique_pred_labels))
+        confusion_matrix = np.zeros((nb_unique_obs_labels,nb_unique_obs_labels))
         
         for observed,predicted in zip(obs_labels,pred_labels):
             confusion_matrix[observed][predicted] += 1
diff --git a/Code/entrainer_tester.py b/Code/entrainer_tester.py
index 4bc132d..f2ef3e7 100644
--- a/Code/entrainer_tester.py
+++ b/Code/entrainer_tester.py
@@ -37,12 +37,9 @@ train, train_labels, test, test_labels = ld.load_iris_dataset(0.7)
 
 myKnn.train(train, train_labels)
 
-for i in range(train.shape[0]):
-    myKnn.predict(train[i],train_labels[i])
-
 # Tester votre classifieur
 
-
+myKnn.test(test, test_labels)