Have crossval tests use predict_partial_hazard

Since other predict methods are numerically unstable atm, use
predict_partial_hazard in k_fold_crossval tests. Since this method
works, I could remove the outer loop and increase expected result
because results are consistently good now (at .93 to 0.97).

Signed-off-by: Jonas Kalderstam <jonas@kalderstam.se>

lifelines/tests/test_suite.py

@@ -1043,33 +1043,39 @@ def test_crossval_for_cox_ph_with_normalizing_times(self):
             times /= np.std(times)
             data_norm['t'] = times
 
-            mean_scores = []
-            for repeat in range(20):
-                scores = k_fold_cross_validation(cf, data_norm,
-                                                duration_col='t',
-                                                event_col='E', k=3)
-                mean_scores.append(np.mean(scores))
+            scores = k_fold_cross_validation(cf, data_norm,
+                                             duration_col='t',
+                                             event_col='E', k=3,
+                                             predictor='predict_partial_hazard')
 
-            expected = 0.85
+            mean_score = np.mean(scores)
+            # partial_hazard will get inverse concordance
+            if mean_score < 0.5:
+                mean_score = 1 - mean_score
+
+            expected = 0.9
             msg = "Expected min-mean c-index {:.2f} < {:.2f}"
-            self.assertTrue(np.mean(mean_scores) > expected,
-                            msg.format(expected, np.mean(mean_scores)))
+            self.assertTrue(mean_score > expected,
+                            msg.format(expected, mean_score))
 
     def test_crossval_for_cox_ph(self):
         cf = CoxPHFitter()
 
         for data_pred in [data_pred1, data_pred2]:
-            mean_scores = []
-            for repeat in range(20):
-                scores = k_fold_cross_validation(cf, data_pred,
-                                                 duration_col='t',
-                                                 event_col='E', k=3)
-                mean_scores.append(np.mean(scores))
+            scores = k_fold_cross_validation(cf, data_pred,
+                                             duration_col='t',
+                                             event_col='E', k=3,
+                                             predictor='predict_partial_hazard')
 
-            expected = 0.85
+            mean_score = np.mean(scores)
+            # partial_hazard will get inverse concordance
+            if mean_score < 0.5:
+                mean_score = 1 - mean_score
+
+            expected = 0.9
             msg = "Expected min-mean c-index {:.2f} < {:.2f}"
-            self.assertTrue(np.mean(mean_scores) > expected,
-                            msg.format(expected, np.mean(mean_scores)))
+            self.assertTrue(mean_score > expected,
+                            msg.format(expected, mean_score))
 
     def test_crossval_for_cox_ph_normalized(self):
         cf = CoxPHFitter()
@@ -1093,16 +1099,19 @@ def test_crossval_for_cox_ph_normalized(self):
                 x2 /= np.std(x2)
                 data_norm['x2'] = x2
 
-            mean_scores = []
-            for repeat in range(20):
-                scores = k_fold_cross_validation(cf, data_norm,
-                                                 duration_col='t',
-                                                 event_col='E', k=3)
-                mean_scores.append(np.mean(scores))
-            expected = 0.85
+            scores = k_fold_cross_validation(cf, data_norm,
+                                             duration_col='t',
+                                             event_col='E', k=3,
+                                             predictor='predict_partial_hazard')
+
+            mean_score = np.mean(scores)
+            # partial_hazard will get inverse concordance
+            if mean_score < 0.5:
+                mean_score = 1 - mean_score
+            expected = 0.9
             msg = "Expected min-mean c-index {:.2f} < {:.2f}"
-            self.assertTrue(np.mean(mean_scores) > expected,
-                            msg.format(expected, np.mean(mean_scores)))
+            self.assertTrue(mean_score > expected,
+                            msg.format(expected, mean_score))
 
     def test_output_against_R(self):
         # from http://cran.r-project.org/doc/contrib/Fox-Companion/appendix-cox-regression.pdf