Estimator

.gitignore

@@ -10,7 +10,7 @@ joblib
 *.pyc
 __pycache__
 *.egg-info
-.coverage
+.coverage*
 
 # IDE specific folders
 .vscode

hidimstat/__init__.py

@@ -1,7 +1,6 @@
 from .adaptive_permutation_threshold import ada_svr
 from .clustered_inference import clustered_inference, hd_inference
 from .desparsified_lasso import desparsified_group_lasso, desparsified_lasso
-from .Dnn_learner_single import DnnLearnerSingle
 from .ensemble_clustered_inference import ensemble_clustered_inference
 from .knockoff_aggregation import knockoff_aggregation
 from .knockoffs import model_x_knockoff
@@ -23,7 +22,6 @@
     "dcrt_zero",
     "desparsified_lasso",
     "desparsified_group_lasso",
-    "DnnLearnerSingle",
     "ensemble_clustered_inference",
     "group_reid",
     "hd_inference",

hidimstat/Dnn_learner_single.py → hidimstat/estimator/Dnn_learner_single.py

@@ -8,7 +8,7 @@
 from sklearn.preprocessing import OneHotEncoder
 from sklearn.utils.validation import check_is_fitted
 
-from .utils import (
+from ._utils.u_Dnn_learner import (
     create_X_y,
     dnn_net,
     joblib_ensemble_dnnet,
@@ -241,6 +241,7 @@ def fit(self, X, y=None):
         loss = np.array(res_ens[4])
 
         if self.n_ensemble == 1:
+            raise Warning("The model can't be fit with n_ensemble = 1")
             return [(res_ens[0][0], (res_ens[1][0], res_ens[2][0]))]
 
         # Keeping the optimal subset of DNNs
@@ -283,6 +284,9 @@ def encode_outcome(self, y, train=True):
             y = y.reshape(-1, 1)
         if self.problem_type == "regression":
             list_y.append(y)
+        # Encoding the target with the ordinal case
+        if self.problem_type == "ordinal":
+            list_y = ordinal_encode(y)
 
         for col in range(y.shape[1]):
             if train:
@@ -291,18 +295,12 @@ def encode_outcome(self, y, train=True):
                     self.enc_y.append(OneHotEncoder(handle_unknown="ignore"))
                     curr_y = self.enc_y[col].fit_transform(y[:, [col]]).toarray()
                     list_y.append(curr_y)
-
-                # Encoding the target with the ordinal case
-                if self.problem_type == "ordinal":
-                    y = ordinal_encode(y)
-
             else:
                 # Encoding the target with the classification case
                 if self.problem_type in ("classification", "binary"):
                     curr_y = self.enc_y[col].transform(y[:, [col]]).toarray()
                     list_y.append(curr_y)
-
-                ## ToDo Add the ordinal case
+
         return np.array(list_y)
 
     def hyper_tuning(

hidimstat/RandomForestModified.py → hidimstat/estimator/RandomForestModified.py

@@ -5,13 +5,13 @@
 class RandomForestClassifierModified(RandomForestClassifier):
     def fit(self, X, y):
         self.y_ = y
-        super().fit(X, y)
+        return super().fit(X, y)
 
     def predict(self, X):
-        super().predict(X)
+        return super().predict(X)
 
     def predict_proba(self, X):
-        super().predict_proba(X)
+        return super().predict_proba(X)
 
     def sample_same_leaf(self, X, y=None):
         if not (y is None):
@@ -42,23 +42,24 @@ def sample_same_leaf(self, X, y=None):
                 )[0]
 
                 # Append the samples to the list
-                leaf_samples.append(
-                    y_minus_i[rng.choice(samples_in_leaf, size=random_samples)]
-                )
+                if samples_in_leaf.size > 0:
+                    leaf_samples.append(
+                        y_minus_i[rng.choice(samples_in_leaf, size=random_samples)]
+                    )
 
             predictions.append(leaf_samples)
 
         # Combine the predictions from all trees to make the final prediction
-        return np.array(predictions)
+        return np.array(predictions, dtype=object)
 
 
 class RandomForestRegressorModified(RandomForestRegressor):
     def fit(self, X, y):
         self.y_ = y
-        super().fit(X, y)
+        return super().fit(X, y)
 
     def predict(self, X):
-        super().predict(X)
+        return super().predict(X)
 
     def sample_same_leaf(self, X):
         rng = np.random.RandomState(self.get_params()["random_state"])
@@ -87,11 +88,12 @@ def sample_same_leaf(self, X):
                 )[0]
 
                 # Append the samples to the list
-                leaf_samples.append(
-                    y_minus_i[rng.choice(samples_in_leaf, size=random_samples)]
-                )
+                if samples_in_leaf.size > 0:
+                    leaf_samples.append(
+                        y_minus_i[rng.choice(samples_in_leaf, size=random_samples)]
+                    )
 
             predictions.append(leaf_samples)
 
         # Combine the predictions from all trees to make the final prediction
-        return np.array(predictions)
+        return np.array(predictions, dtype=object)

hidimstat/estimator/__init__.py

@@ -0,0 +1,5 @@
+from .Dnn_learner_single import DnnLearnerSingle
+
+__all__ = [
+    "DnnLearnerSingle",
+]