API + fix unit test on regression

sdpython · Feb 10, 2018 · cad53f3 · cad53f3
1 parent d7086e6
commit cad53f3
Show file tree

Hide file tree

Showing 5 changed files with 28 additions and 6 deletions.
diff --git a/_doc/sphinxdoc/source/api/datasets.rst b/_doc/sphinxdoc/source/api/datasets.rst
@@ -6,7 +6,12 @@ Jeux de données
 .. contents::
     :local:
 
+Recommandations
+===============
+
+.. autosignature:: papierstat.datasets.load_movielens_dataset
+
 Régression
-++++++++++
+==========
 
 .. autosignature:: papierstat.datasets.load_wines_dataset
diff --git a/_doc/sphinxdoc/source/api/index.rst b/_doc/sphinxdoc/source/api/index.rst
@@ -7,3 +7,4 @@ API
     :maxdepth: 1
 
     datasets
+    mltricks
diff --git a/_doc/sphinxdoc/source/api/mltricks.rst b/_doc/sphinxdoc/source/api/mltricks.rst
@@ -0,0 +1,12 @@
+
+===========================
+Astuces de machine learning
+===========================
+
+.. contents::
+    :local:
+
+Autour de scikit-learn
+======================
+
+.. autosignature:: papierstat.mltricks.sklearn_base_transform_learner.SkBaseTransformLearner
diff --git a/_unittests/ut_mltricks/test_sklearn_convert.py b/_unittests/ut_mltricks/test_sklearn_convert.py
@@ -42,7 +42,7 @@
 from sklearn.datasets import load_iris
 from sklearn.linear_model import LogisticRegression, LinearRegression
 from sklearn.tree import DecisionTreeClassifier, DecisionTreeRegressor
-from sklearn.metrics import accuracy_score
+from sklearn.metrics import accuracy_score, r2_score
 from sklearn.pipeline import make_pipeline
 from src.papierstat.mltricks import SkBaseTransformLearner
 
@@ -105,13 +105,13 @@ def test_pipeline_with_two_regressors(self):
         pipe = make_pipeline(conv, DecisionTreeRegressor())
         pipe.fit(X_train, y_train)
         pred = pipe.predict(X_test)
-        score = accuracy_score(y_test, pred)
-        self.assertGreater(score, 0.92)
+        score = r2_score(y_test, pred)
+        self.assertLesser(score, 1.)
         score2 = pipe.score(X_test, y_test)
         self.assertEqual(score, score2)
         rp = repr(conv)
         self.assertStartsWith(
-            'SkBaseTransformLearner(model=LogisticRegression(C=1.0,', rp)
+            'SkBaseTransformLearner(model=LinearRegression(copy_X=True,', rp)
 
 
 if __name__ == "__main__":

diff --git a/src/papierstat/mltricks/sklearn_base_transform_learner.py b/src/papierstat/mltricks/sklearn_base_transform_learner.py
@@ -4,6 +4,7 @@
 @brief Implémente un *transform* qui suit la même API que tout :epkg:`scikit-learn` transform.
 """
 import textwrap
+import numpy
 from .sklearn_base_transform import SkBaseTransform
 
 
@@ -110,7 +111,10 @@ def transform(self, X):
         @param      X   features
         @return         prédictions
         """
-        return self.method(X)
+        res = self.method(X)
+        if len(res.shape) == 1:
+            res = res[:, numpy.newaxis]
+        return res
 
     ##############
     # cloning API