Add files via upload

classification.py

@@ -174,6 +174,9 @@ def setup(data,
     X = data.drop(target,axis=1)
     y = data[target]
     
+    #copy original data for pandas profiler
+    data_before_preprocess = data.copy()
+    
     #determining target type
     if y.value_counts().count() > 2:
         target_type = 'Multiclass'
@@ -345,7 +348,11 @@ def setup(data,
             '''
             clear_output()
             print(' ')
-            print('Setup Succesfully Completed!')
+            if profile:
+                print('Setup Succesfully Completed! Loading Profile Now... Please Wait!')
+            else:
+                print('Setup Succesfully Completed!')
+            
             functions = pd.DataFrame ( [ ['session_id', seed ],
                                          ['Target Type', target_type],
                                          ['Original Data',X.shape ], 
@@ -358,6 +365,15 @@ def setup(data,
             functions_ = functions.style.hide_index()
             display(functions_)
             
+            if profile:
+                try:
+                    import pandas_profiling
+                    pf = pandas_profiling.ProfileReport(data_before_preprocess)
+                    clear_output()
+                    display(pf)
+                except:
+                    print('Data Profiler Failed. No output to show, please continue with Modeling.')
+            
             '''
             Final display Ends
             '''   
@@ -389,7 +405,11 @@ def setup(data,
                 
             clear_output()
             print(' ')
-            print('Setup Succesfully Completed!')
+            if profile:
+                print('Setup Succesfully Completed! Loading Profile Now... Please Wait!')
+            else:
+                print('Setup Succesfully Completed!')
+                
             functions = pd.DataFrame ( [ ['session_id', seed ],
                                          ['Target Type', target_type],
                                          ['Original Data',X.shape ], 
@@ -402,6 +422,15 @@ def setup(data,
             functions_ = functions.style.hide_index()
             display(functions_)
             
+            if profile:
+                try:
+                    import pandas_profiling
+                    pf = pandas_profiling.ProfileReport(data_before_preprocess)
+                    clear_output()
+                    display(pf)
+                except:
+                    print('Data Profiler Failed. No output to show, please continue with Modeling.')
+            
             '''
             Final display Ends
             ''' 
@@ -446,10 +475,13 @@ def setup(data,
         display(functions_)
         
         if profile:
-            import pandas_profiling
-            pf = pandas_profiling.ProfileReport(data)
-            clear_output()
-            display(pf)
+            try:
+                import pandas_profiling
+                pf = pandas_profiling.ProfileReport(data_before_preprocess)
+                clear_output()
+                display(pf)
+            except:
+                print('Data Profiler Failed. No output to show, please continue with Modeling.')
             
         '''
         Final display Ends
@@ -3481,6 +3513,7 @@ def tune_model(estimator = None,
         return best_model
 
 
+
 def blend_models(estimator_list = 'All', 
                  fold = 10, 
                  round = 4, 
@@ -3591,6 +3624,21 @@ def blend_models(estimator_list = 'All',
             if 'sklearn' not in str(type(i)) and 'CatBoostClassifier' not in str(type(i)):
                 sys.exit("(Value Error): estimator_list parameter only accepts 'All' as string or trained model object")
    
+    #checking method param with estimator list
+    if estimator_list != 'All':
+        if method == 'soft':
+            
+            check = 0
+            
+            for i in estimator_list:
+                if hasattr(i, 'predict_proba'):
+                    pass
+                else:
+                    check += 1
+            
+            if check >= 1:
+                sys.exit('(Type Error): Estimator list contains estimator that doesnt support probabilities and method is forced to soft. Either change the method or drop the estimator.')
+    
     #checking fold parameter
     if type(fold) is not int:
         sys.exit('(Type Error): Fold parameter only accepts integer value.')
@@ -3743,12 +3791,20 @@ def blend_models(estimator_list = 'All',
         progress.value += 1
         
         if turbo:
-            estimator_list = [lr,knn,nb,dt,svm,ridge,rf,qda,ada,gbc,lda,et,xgboost,lightgbm]
-            voting = 'hard'
+            if method == 'hard':
+                estimator_list = [lr,knn,nb,dt,svm,ridge,rf,qda,ada,gbc,lda,et,xgboost,lightgbm]
+                voting = 'hard'
+            elif method == 'soft':
+                estimator_list = [lr,knn,nb,dt,rf,qda,ada,gbc,lda,et,xgboost,lightgbm]
+                voting = 'soft'
         else:
-            estimator_list = [lr,knn,nb,dt,svm,rbfsvm,gpc,mlp,ridge,rf,qda,ada,gbc,lda,et,xgboost,lightgbm]
-            voting = 'hard'
-
+            if method == 'hard':
+                estimator_list = [lr,knn,nb,dt,svm,rbfsvm,gpc,mlp,ridge,rf,qda,ada,gbc,lda,et,xgboost,lightgbm]
+                voting = 'hard'
+            elif method == 'soft':
+                estimator_list = [lr,knn,nb,dt,rbfsvm,gpc,mlp,rf,qda,ada,gbc,lda,et,xgboost,lightgbm]
+                voting = 'soft'
+                
     else:
 
         estimator_list = estimator_list

clustering.py

@@ -387,17 +387,12 @@ def create_model(model = None,
         
     elif model == 'kmodes':
         from kmodes.kmodes import KModes
-        model = KModes(n_clusters=num_clusters, n_jobs=-1, random_state=seed)
+        model = KModes(n_clusters=num_clusters, n_jobs=1, random_state=seed)
         full_name = 'K-Modes Clustering'
         
-    elif model == 'kprototypes':
-        from kmodes.kprototypes import KPrototypes
-        model = KPrototypes(n_clusters=num_clusters, n_jobs=-1, random_state=seed)
-        full_name = 'K-Prototypes Clustering'
-        
     elif model == 'skmeans':
         from spherecluster import SphericalKMeans
-        model = SphericalKMeans(n_clusters=num_clusters, n_jobs=-1, random_state=seed)
+        model = SphericalKMeans(n_clusters=num_clusters, n_jobs=1, random_state=seed)
         full_name = 'Spherical K-Means Clustering'
         
     #monitor update

regression.py

@@ -163,6 +163,9 @@ def setup(data,
     X = data.drop(target,axis=1)
     y = data[target]
     
+    #copy original data for pandas profiler
+    data_before_preprocess = data.copy()
+    
     progress.value += 1
     
     if sampling is True and data.shape[0] > 25000: #change back to 25000
@@ -271,7 +274,12 @@ def setup(data,
             '''
             clear_output()
             print(' ')
-            print('Setup Succesfully Completed!')
+
+            if profile:
+                print('Setup Succesfully Completed! Loading Profile Now... Please Wait!')
+            else:
+                print('Setup Succesfully Completed!')    
+        
             functions = pd.DataFrame ( [ ['session_id', seed ],
                                          ['Original Data',X.shape ], 
                                          ['Sampled Data',X.shape ], 
@@ -283,6 +291,15 @@ def setup(data,
             functions_ = functions.style.hide_index()
             display(functions_)
             
+            if profile:
+                try:
+                    import pandas_profiling
+                    pf = pandas_profiling.ProfileReport(data_before_preprocess)
+                    clear_output()
+                    display(pf)
+                except:
+                    print('Data Profiler Failed. No output to show, please continue with Modeling.')
+            
             '''
             Final display Ends
             '''   
@@ -313,7 +330,12 @@ def setup(data,
             
             clear_output()
             print(' ')
-            print('Setup Succesfully Completed!')
+            
+            if profile:
+                print('Setup Succesfully Completed! Loading Profile Now... Please Wait!')
+            else:
+                print('Setup Succesfully Completed!')
+            
             functions = pd.DataFrame ( [ ['session_id', seed ],
                                          ['Original Data',X.shape ], 
                                          ['Sampled Data',X_selected.shape ], 
@@ -325,6 +347,15 @@ def setup(data,
             functions_ = functions.style.hide_index()
             display(functions_)
             
+            if profile:
+                try:
+                    import pandas_profiling
+                    pf = pandas_profiling.ProfileReport(data_before_preprocess)
+                    clear_output()
+                    display(pf)
+                except:
+                    print('Data Profiler Failed. No output to show, please continue with Modeling.')
+            
             '''
             Final display Ends
             ''' 
@@ -368,10 +399,13 @@ def setup(data,
         display(functions_)
             
         if profile:
-            import pandas_profiling
-            pf = pandas_profiling.ProfileReport(data)
-            clear_output()
-            display(pf)
+            try:
+                import pandas_profiling
+                pf = pandas_profiling.ProfileReport(data_before_preprocess)
+                clear_output()
+                display(pf)
+            except:
+                print('Data Profiler Failed. No output to show, please continue with Modeling.')
             
         '''
         Final display Ends
@@ -386,6 +420,8 @@ def setup(data,
         
         return X, y, X_train, X_test, y_train, y_test, seed, experiment__
 
+
+
 def create_model(estimator = None, 
                  ensemble = False, 
                  method = None,