wip: multitask api

Author: mrahim

examples/multitask_example.py

@@ -0,0 +1,21 @@
+import numpy as np
+from sklearn.datasets import make_classification
+from stlearn import MultiTaskEstimator
+from sklearn.linear_model import (MultiTaskLassoCV,
+                                  MultiTaskElasticNet,
+                                  MultiTaskElasticNetCV,
+                                  LassoCV, LogisticRegression)
+from sklearn.metrics import accuracy_score
+
+X, y = make_classification(random_state=25)
+n = 10
+
+Y = np.array(n*[y]).T
+
+mt = MultiTaskEstimator(
+                        # estimator=MultiTaskLassoCV(alphas=np.logspace(-3, 3, 7)),
+                        estimator=MultiTaskElasticNetCV(),
+                        output_types=n*['binary'])
+
+mt.fit(X, Y)
+print(mt.score(X, Y))

examples/fmri_stacking.py renamed to examples/stacking_example.py

@@ -10,7 +10,8 @@
 from sklearn.base import BaseEstimator, TransformerMixin
 from sklearn.metrics import accuracy_score, r2_score
 from sklearn.externals.joblib import Memory, Parallel, delayed
-from sklearn.linear_model import MultiTaskLassoCV, MultiTaskElasticNetCV
+from sklearn.linear_model import (
+    MultiTaskLassoCV, MultiTaskElasticNetCV, LogisticRegression)
 
 
 class MultiTaskEstimator(BaseEstimator, TransformerMixin):
@@ -44,12 +45,12 @@ def _decision_function(self, X):
 
     def predict(self, X):
         # predict multiple outputs
-        Ypred = self.estimator._decision_function(X)
+        Ypred = self._decision_function(X)
         for i in range(self.n_outputs):
             if self.output_types[i] == 'binary':
-                # binarize
+                # binarize classification results
                 labels = np.zeros(Ypred[:, i].shape)
-                labels[Ypred[:, i] > 0.] = 1
+                labels[Ypred[:, i] >= 0.5] = 1
                 Ypred[:, i] = labels
         return Ypred