Merge pull request #7 from andrewtavis/add-testing

Tests for evaluation metrics

tests/conftest.py

@@ -400,6 +400,11 @@ def df_vis_eval_proba(request):
     return request.param
 
 
+@pytest.fixture(params=[model_eval_dict_proba])
+def model_evaluation_dict_proba(request):
+    return request.param
+
+
 # Iterated pred models
 tm = two_model.TwoModel(
     treatment_model=RandomForestRegressor(random_state=42),

tests/test_evaluation.py

@@ -12,61 +12,86 @@
 
 np.random.seed(42)
 
+models = ["two_model", "interaction_term"]
 
-def test_plot_cum_effect(monkeypatch):
-    monkeypatch.setattr(plt, "show", lambda: None)
-    # evaluation.get_cum_effect(
-    #     df,
-    #     models=None,
-    #     outcome_col="y",
-    #     treatment_col="w",
-    #     treatment_effect_col="tau",
-    #     normalize=False,
-    #     random_seed=None,
-    # )
-    assert True
-
-
-def test_plot_cum_gain(monkeypatch):
-    monkeypatch.setattr(plt, "show", lambda: None)
-    assert True
 
-
-def test_plot_qini(monkeypatch):
+def test_plot_cum_effect(monkeypatch, df_vis_eval_proba):
     monkeypatch.setattr(plt, "show", lambda: None)
-    assert True
-
-
-def test_auuc_score():
-    assert True
-
-
-def test_qini_score():
-    assert True
-
-
-def test_get_batch_metrics():
-    assert True
-
-
-def test_plot_batch_metrics(monkeypatch):
+    evaluation.plot_cum_effect(
+        df=df_vis_eval_proba,
+        n=20,
+        models=models,
+        percent_of_pop=False,
+        outcome_col="y_test",
+        treatment_col="w_test",
+        random_seed=42,
+        figsize=(10, 5),
+        fontsize=20,
+        axis=None,
+        legend_metrics=False,
+    )
+
+
+def test_plot_cum_gain(monkeypatch, df_vis_eval_proba):
     monkeypatch.setattr(plt, "show", lambda: None)
-    assert True
-
-
-def test_plot_batch_responses(monkeypatch):
+    evaluation.plot_cum_gain(
+        df=df_vis_eval_proba,
+        n=100,
+        models=models,
+        percent_of_pop=True,
+        outcome_col="y_test",
+        treatment_col="w_test",
+        normalize=True,
+        random_seed=42,
+        figsize=None,
+        fontsize=20,
+        axis=None,
+        legend_metrics=True,
+    )
+
+
+def test_plot_qini(monkeypatch, df_vis_eval_proba):
+    monkeypatch.setattr(plt, "show", lambda: None)
+    evaluation.plot_qini(
+        df=df_vis_eval_proba,
+        n=100,
+        models=models,
+        percent_of_pop=True,
+        outcome_col="y_test",
+        treatment_col="w_test",
+        normalize=True,
+        random_seed=42,
+        figsize=None,
+        fontsize=20,
+        axis=None,
+        legend_metrics=True,
+    )
+
+
+def test_plot_batch_responses(monkeypatch, df_vis_eval_proba):
     monkeypatch.setattr(plt, "show", lambda: None)
-    assert True
+    evaluation.plot_batch_responses(
+        df=df_vis_eval_proba,
+        n=10,
+        models=models,
+        outcome_col="y_test",
+        treatment_col="w_test",
+        normalize=False,
+        figsize=None,
+        fontsize=15,
+        axis=None,
+    )
 
 
 def test_signal_to_noise(y_split_proba, w_split_proba):
     sn_ration = evaluation.signal_to_noise(y=y_split_proba, w=w_split_proba)
     assert type(sn_ration) == float or type(sn_ration) == np.float64
 
 
-def test_iterate_model_pred():
-    assert True
-
-
-def test_pred_proba_eval_table():
-    assert True
+def test_pred_proba_eval_table(model_evaluation_dict_proba):
+    evaluation.eval_table(
+        model_evaluation_dict_proba, variances=True, annotate_vars=True
+    )
+    evaluation.eval_table(
+        model_evaluation_dict_proba, variances=False, annotate_vars=False
+    )