fix classifiers -- bump v0.36.0

Author: thierrymoudiki

examples/ridge.py

@@ -10,37 +10,39 @@
 
 # Single lambda
 print("\n\n Single lambda")
-model = ns.RidgeRegressor(lambda_=1.0)
+model = ns.RidgeRegressor(reg_lambda=1.0)
 start = time()
 model.fit(X_train, y_train)
 predictions = model.predict(X_test)
 end = time()
 print(f"Time taken: {end - start} seconds")
 print(f"coefs: {model.coef_}")
+print(f"Test RMSE: {np.sqrt(np.mean((predictions - y_test)**2))}")
+
 # Access criteria
-print(f"GCV scores: {model.GCV_}")
-print(f"HKB estimate: {model.HKB_}")
-print(f"LW estimate: {model.LW_}")
+#print(f"GCV scores: {model.GCV_}")
+#print(f"HKB estimate: {model.HKB_}")
+#print(f"LW estimate: {model.LW_}")
 
 
 # Multiple lambdas
-print("\n\n Multiple lambdas")
-model = ns.RidgeRegressor(lambda_=[0.1, 1.0, 10.0])
-start = time()
-model.fit(X_train, y_train)
-end = time()
-print(f"Time taken: {end - start} seconds")
-predictions = model.predict(X_test)  # Returns predictions for each lambda
-print(f"coefs: {model.coef_}")
-# Access criteria
-print(f"GCV scores: {model.GCV_}")
-print(f"HKB estimate: {model.HKB_}")
-print(f"LW estimate: {model.LW_}")
+# print("\n\n Multiple lambdas")
+# model = ns.RidgeRegressor(lambda_=[0.1, 1.0, 10.0])
+# start = time()
+# model.fit(X_train, y_train)
+# end = time()
+# print(f"Time taken: {end - start} seconds")
+# predictions = model.predict(X_test)  # Returns predictions for each lambda
+# print(f"coefs: {model.coef_}")
+# # Access criteria
+# print(f"GCV scores: {model.GCV_}")
+# print(f"HKB estimate: {model.HKB_}")
+# print(f"LW estimate: {model.LW_}")
 
-# Compare coefficients for lambda=1.0
-print("\nComparing coefficients for lambda=1.0:")
-print("Single lambda coefs:", model.coef_)
-print("Multiple lambda coefs (middle column):", model.coef_[:, 1])
+# # Compare coefficients for lambda=1.0
+# print("\nComparing coefficients for lambda=1.0:")
+# print("Single lambda coefs:", model.coef_)
+# print("Multiple lambda coefs (middle column):", model.coef_[:, 1])
 
-# With GPU acceleration
-#model = RidgeRegressor(lambda_=1.0, backend="gpu")
+# # With GPU acceleration
+# #model = RidgeRegressor(lambda_=1.0, backend="gpu")

nnetsauce/boosting/adaBoostClassifier.py

@@ -155,6 +155,7 @@ class AdaBoostClassifier(Boosting, ClassifierMixin):
     """
 
     # construct the object -----
+    _estimator_type = "classifier"
 
     def __init__(
         self,
@@ -461,3 +462,8 @@ def predict_proba(self, X, **kwargs):
         sum_ensemble = expit_ensemble_learner.sum(axis=1)
 
         return expit_ensemble_learner / sum_ensemble[:, None]
+
+    @property
+    def _estimator_type(self):
+        return "classifier"            
+

nnetsauce/custom/customClassifier.py

@@ -137,6 +137,7 @@ class CustomClassifier(Custom, ClassifierMixin):
     """
 
     # construct the object -----
+    _estimator_type = "classifier" 
 
     def __init__(
         self,
@@ -182,7 +183,6 @@ def __init__(
         self.type_fit = "classification"
         self.cv_calibration = cv_calibration
         self.calibration_method = calibration_method
-        self._estimator_type = "classifier"  # Explicitly mark as classifier
 
     def __sklearn_clone__(self):
         """Create a clone of the estimator.
@@ -478,4 +478,8 @@ def score(self, X, y, scoring=None):
             return -skm2.brier_score_loss(y, self.predict_proba(X))
 
         if scoring == "neg_log_loss":
-            return -skm2.log_loss(y, self.predict_proba(X))
+            return -skm2.log_loss(y, self.predict_proba(X))
+
+    @property
+    def _estimator_type(self):
+        return "classifier"

nnetsauce/deep/deepClassifier.py

@@ -65,6 +65,7 @@ class DeepClassifier(CustomClassifier, ClassifierMixin):
         print(clf.score(clf.predict(X_test), y_test))
         ```
     """
+    _estimator_type = "classifier"
 
     def __init__(
         self,
@@ -129,7 +130,6 @@ def __init__(
         self.cv_calibration = cv_calibration
         self.calibration_method = calibration_method
         self.obj = obj
-        self._estimator_type = "classifier"  # Add this line to explicitly mark as classifier
 
         assert n_layers >= 1, "must have n_layers >= 1"
         self.stacked_obj = obj
@@ -636,3 +636,7 @@ def lazy_cross_val_optim(
                         pass
 
         return results
+
+    @property
+    def _estimator_type(self):
+        return "classifier"        

nnetsauce/glm/glmClassifier.py

@@ -89,6 +89,7 @@ class GLMClassifier(GLM, ClassifierMixin):
     """
 
     # construct the object -----
+    _estimator_type = "classifier"
 
     def __init__(
         self,
@@ -377,3 +378,7 @@ def score(self, X, y, scoring=None):
 
         if scoring == "neg_log_loss":
             return -skm2.log_loss(y, self.predict_proba(X))
+
+    @property
+    def _estimator_type(self):
+        return "classifier"            

nnetsauce/multitask/multitaskClassifier.py

@@ -124,6 +124,7 @@ class MultitaskClassifier(Base, ClassifierMixin):
     """
 
     # construct the object -----
+    _estimator_type = "classifier"
 
     def __init__(
         self,
@@ -306,3 +307,8 @@ def decision_function(self, X, **kwargs):
             )[0]
 
         return self.obj.decision_function(self.cook_test_set(X, **kwargs), **kwargs)
+
+    @property
+    def _estimator_type(self):
+        return "classifier"            
+        

nnetsauce/multitask/simplemultitaskClassifier.py

@@ -72,6 +72,7 @@ class SimpleMultitaskClassifier(Base, ClassifierMixin):
     """
 
     # construct the object -----
+    _estimator_type = "classifier"
 
     def __init__(
         self,
@@ -232,3 +233,7 @@ def decision_function(self, X, **kwargs):
             )[0]
 
         return self.obj.decision_function(self.cook_test_set(X, **kwargs), **kwargs)
+
+    @property
+    def _estimator_type(self):
+        return "classifier"            

nnetsauce/neuralnet/neuralnetclassification.py

@@ -55,6 +55,7 @@ class NeuralNetClassifier(BaseEstimator, ClassifierMixin):
         set_weights(weights)
             Set the weights of the model.
     """
+    _estimator_type = "classifier"
 
     def __init__(
         self,
@@ -137,4 +138,8 @@ def predict(self, X):
                 Training vectors, where n_samples is the number of samples and
                 n_features is the number of features.
         """
-        return self.regr.predict(X)
+        return self.regr.predict(X)
+
+    @property
+    def _estimator_type(self):
+        return "classifier"            

nnetsauce/randombag/randomBagClassifier.py

@@ -129,6 +129,7 @@ class RandomBagClassifier(RandomBag, ClassifierMixin):
     """
 
     # construct the object -----
+    _estimator_type = "classifier"
 
     def __init__(
         self,
@@ -373,3 +374,9 @@ def predict_estimator(m):
             ensemble_proba += weights[i] * preds[i]
 
         return ensemble_proba
+
+
+    @property
+    def _estimator_type(self):
+        return "classifier"            
+