Merge pull request #28 from maskani-moh/pep8-fix

Pep8 fix

python-package/mlbox/preprocessing/__init__.py

@@ -2,4 +2,4 @@
 from .reader import *
 
 import time
-time.sleep(30) #waiting for the engines to start
+time.sleep(30)  # Waiting for the engines to start

python-package/mlbox/preprocessing/drift/__init__.py

@@ -6,4 +6,6 @@
 import os
 os.system("ipcluster start --profile=home &")
 
-warnings.warn("ipCluster is starting. Please wait 30 sec and check in terminal that 'the engines appear to have started successfully'.")
+warnings.warn("ipCluster is starting. "
+              "Please wait 30 sec and check in terminal that "
+              "'the engines appear to have started successfully'.")

python-package/mlbox/preprocessing/drift/drift_estimator.py

@@ -9,8 +9,8 @@
 from sklearn.metrics import roc_auc_score
 from sklearn.model_selection import KFold, StratifiedKFold
 
+
 def cross_val_predict_proba(estimator, X, y, cv):
-
 
     """Evaluates the target by cross-validation
 
@@ -35,21 +35,19 @@ def cross_val_predict_proba(estimator, X, y, cv):
 
     """
 
-
     y_pred = np.zeros(len(y))
-    
-    for train_index, test_index in cv.split(X,y):
-        
+
+    for train_index, test_index in cv.split(X, y):
+
         estimator.fit(X.iloc[train_index], y.iloc[train_index])
-
-        y_pred[test_index] = estimator.predict_proba(X.iloc[test_index])[:,1]
-
-    return y_pred
 
+        y_pred[test_index] = estimator.predict_proba(X.iloc[test_index])[:, 1]
+
+    return y_pred
 
 
 class DriftEstimator():
-    
+
     """Estimates the drift between two datasets
     
         
@@ -60,35 +58,42 @@ class DriftEstimator():
         
     n_folds : int, defaut = 2
         Number of folds used to estimate the drift
-        
+
     stratify : bool, defaut = True
         Whether the cv is stratified (same number of train and test samples within each fold)
 
     random_state : int, defaut = 1
         Random state for cv
     """
-
-    def __init__(self, estimator = RandomForestClassifier(n_estimators = 50, n_jobs=-1, max_features=1., min_samples_leaf = 5, max_depth = 5), n_folds = 2, stratify = True, random_state = 1):
+
+    def __init__(self,
+                 estimator=RandomForestClassifier(n_estimators=50,
+                                                  n_jobs=-1,
+                                                  max_features=1.,
+                                                  min_samples_leaf=5,
+                                                  max_depth=5),
+                 n_folds=2,
+                 stratify=True,
+                 random_state=1):
 
         self.estimator = estimator
         self.n_folds = n_folds
         self.stratify = stratify
         self.random_state = random_state
         self.__cv = None
         self.__pred = None
+        # TODO: Change to 'cible' to english 'target'?
         self.__cible = None
-        self.__fitOK = False                                   
-
-
+        self.__fitOK = False
+
     def get_params(self):
 
-        return {'estimator': self.estimator, 
-            'n_folds' : self.n_folds,
-            'stratify' : self.stratify,
-            'random_state' : self.random_state}
+        return {'estimator': self.estimator,
+                'n_folds': self.n_folds,
+                'stratify': self.stratify,
+                'random_state': self.random_state}
 
-
-    def set_params(self,**params):
+    def set_params(self, **params):
 
         if('estimator' in params.keys()):
             self.estimator = params['estimator']
@@ -97,12 +102,12 @@ def set_params(self,**params):
         if('stratify' in params.keys()):
             self.stratify = params['stratify']
         if('random_state' in params.keys()):
-            self.random_state = params['random_state']       
-
-
+            self.random_state = params['random_state']
+
     def fit(self, df_train, df_test):
-
-        """Computes the drift between the two datasets
+
+        """
+        Computes the drift between the two datasets
 
         Parameters
         ----------
@@ -117,30 +122,37 @@ def fit(self, df_train, df_test):
         self : object
             Returns self.
         """
-        
+
         df_train["target"] = 0
         df_test["target"] = 1
 
-        self.__cible = pd.concat((df_train.target, df_test.target),ignore_index=True)
-
-        if(self.stratify == True):
-            self.__cv = StratifiedKFold(n_splits = self.n_folds, shuffle = True, random_state = self.random_state)
+        self.__cible = pd.concat((df_train.target, df_test.target),
+                                 ignore_index=True)
+
+        if self.stratify:
+            self.__cv = StratifiedKFold(n_splits=self.n_folds,
+                                        shuffle=True,
+                                        random_state=self.random_state)
         else:
-            self.__cv = KFold(n_splits = self.n_folds, shuffle = True, random_state = self.random_state)
+            self.__cv = KFold(n_splits=self.n_folds,
+                              shuffle=True,
+                              random_state=self.random_state)
 
-
-        self.__pred = cross_val_predict_proba(estimator = self.estimator,
-                                              X = pd.concat((df_train, df_test),ignore_index=True).drop(['target'], axis = 1),
-                                              y = self.__cible,
-                                              cv = self.__cv)
+        X_tmp = pd.concat((df_train, df_test),
+                          ignore_index=True).drop(['target'], axis=1)
+
+        self.__pred = cross_val_predict_proba(estimator=self.estimator,
+                                              X=X_tmp,
+                                              y=self.__cible,
+                                              cv=self.__cv)
 
         del df_train["target"]
         del df_test["target"]
+
         self.__fitOK = True
 
         return self
-
-
+
     def score(self):
 
         """Returns the global drift measure between two datasets.
@@ -152,35 +164,38 @@ def score(self):
         float
             The drift measure
         """
-        
+
         S = []
-        
+
         if self.__fitOK:
 
-            for train_index, test_index in self.__cv.split(X=np.zeros(len(self.__cible)), y=self.__cible):
-
-                S.append(roc_auc_score(self.__cible.iloc[test_index], self.__pred[test_index]))
-
-            return max(np.mean(S),1-np.mean(S))   # renvoyer la moyenne des AUC symetrisee
-
+            X_zeros = np.zeros(len(self.__cible))
+
+            for train_index, test_index in self.__cv.split(X=X_zeros,
+                                                           y=self.__cible):
+
+                S.append(roc_auc_score(self.__cible.iloc[test_index],
+                                       self.__pred[test_index]))
+
+            # TODO: return the mean of mirrored AUC metrics
+            return max(np.mean(S), 1-np.mean(S))
+
         else:
-            raise ValueError('Call the fit function before !')  
-
-
-    def predict(self): 
-
-        """Returns the probabilities that the sample belongs to the test dataset 
+            raise ValueError('Call the fit function before !')
+
+    def predict(self):
+
+        """Returns the probabilities that the sample belongs to the test dataset
 
         Returns
         -------
         Array of shape = (n_train+n_test,)
             The probabilities
         """
 
-
         if self.__fitOK:
-            
+
             return self.__pred
-        
+
         else:
             raise ValueError('Call the fit function before !')