Structured test cases

openml · Sep 29, 2016 · b250547 · b250547
1 parent 0f58378
commit b250547
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 15 deletions.
diff --git a/openml/runs/run.py b/openml/runs/run.py
@@ -120,7 +120,6 @@ def _generate_trace_arff_dict(self, model):
         arff_dict['data'] = self.trace_content
         arff_dict['relation'] = 'openml_task_' + str(self.task_id) + '_predictions'
 
-        print(arff_dict)
         return arff_dict
 
     def publish(self):

diff --git a/tests/runs/test_run_functions.py b/tests/runs/test_run_functions.py
@@ -1,38 +1,66 @@
 from sklearn.linear_model import LogisticRegression, SGDClassifier
-from sklearn.ensemble import RandomForestClassifier
-from sklearn.model_selection import RandomizedSearchCV
+from sklearn.ensemble import RandomForestClassifier, BaggingClassifier
+from sklearn.svm import SVC
+from sklearn.model_selection import RandomizedSearchCV, GridSearchCV
 import openml
 from openml.testing import TestBase
 
 
 class TestRun(TestBase):
-    def test_run_iris(self):
-        task = openml.tasks.get_task(10107)
-        clf = LogisticRegression()
+
+    def _perform_run(self, task_id, num_instances, clf):
+        task = openml.tasks.get_task(task_id)
         run = openml.runs.run_task(task, clf)
         run_ = run.publish()
         self.assertEqual(run_, run)
         self.assertIsInstance(run.dataset_id, int)
 
-    def test_run_optimize_randomforest_iris(self):
-        task = openml.tasks.get_task(10107)
-        numIterations = 5
+        # check arff output
+        self.assertEqual(len(run.data_content), num_instances)
+        return run
 
 
-        clf = RandomForestClassifier(n_estimators=numIterations)
+    def test_run_iris(self):
+        task_id = 10107
+        num_instances = 150
 
+        clf = LogisticRegression()
+        self._perform_run(task_id,num_instances, clf)
+
+
+    def test_run_optimize_randomforest_iris(self):
+        task_id = 10107
+        num_instances = 150
+        num_folds = 10
+        num_iterations = 5
+
+        clf = RandomForestClassifier(n_estimators=10)
         param_dist = {"max_depth": [3, None],
                       "max_features": [1,2,3,4],
                       "min_samples_split": [1,2,3,4,5,6,7,8,9,10],
                       "min_samples_leaf": [1,2,3,4,5,6,7,8,9,10],
                       "bootstrap": [True, False],
                       "criterion": ["gini", "entropy"]}
-        random_search = RandomizedSearchCV(clf, param_dist,n_iter=20)
+        random_search = RandomizedSearchCV(clf, param_dist,n_iter=num_iterations)
+
+        run = self._perform_run(task_id,num_instances, random_search)
+        self.assertEqual(len(run.trace_content), num_iterations * num_folds)
+
+    def test_run_optimize_bagging_iris(self):
+        task_id = 10107
+        num_instances = 150
+        num_folds = 10
+        num_iterations = 36 # (num values for C times gamma)
+
+        task = openml.tasks.get_task(task_id)
+        bag = BaggingClassifier(base_estimator=SVC())
+        param_dist = {"base_estimator__C": [0.001, 0.01, 0.1, 1, 10, 100],
+                      "base_estimator__gamma": [0.001, 0.01, 0.1, 1, 10, 100]}
+        grid_search = GridSearchCV(bag, param_dist)
+
+        run = self._perform_run(task_id, num_instances, grid_search)
+        self.assertEqual(len(run.trace_content), num_iterations * num_folds)
 
-        run = openml.runs.run_task(task, random_search)
-        run_ = run.publish()
-        self.assertEqual(run_, run)
-        self.assertIsInstance(run.dataset_id, int)
 
     def test__run_task_get_arffcontent(self):
         task = openml.tasks.get_task(1939)