Merge pull request #29 from chiuhoward/fix-pytorch

[FIX] Remove deprecated verbose flag from ReduceLROnPlateau in tests

Author: arokem

afqinsight/match.py

@@ -4,10 +4,10 @@
 import pandas as pd
 from scipy.optimize import linear_sum_assignment
 
-__all__ = ["mahalonobis_dist_match"]
+__all__ = ["mahalanobis_dist_match"]
 
 
-def mahalonobis_dist_match(
+def mahalanobis_dist_match(
     data=None, test=None, ctrl=None, status_col=None, feature_cols=None, threshold=0.2
 ):
     """
@@ -77,12 +77,12 @@ def mahalonobis_dist_match(
             (
                 "There are NaNs in test or ctrl data. "
                 "Please replace these NaNs using interpolation or by removing "
-                "the subjects with NaNs before calling mahalonobis_dist_match. "
+                "the subjects with NaNs before calling mahalanobis_dist_match. "
             )
         )
 
-    # calculate Mahalonobis distance between test and control
-    nbrs = _mahalonobis_dist(test, ctrl)
+    # calculate Mahalanobis distance between test and control
+    nbrs = _mahalanobis_dist(test, ctrl)
 
     # assign neighbors using Munkres algorithm
     row_ind, col_ind = linear_sum_assignment(nbrs)
@@ -118,9 +118,9 @@ def mahalonobis_dist_match(
         return data.iloc[all_idx]
 
 
-def _mahalonobis_dist(arr1, arr2):
+def _mahalanobis_dist(arr1, arr2):
     """
-    Calculate the Mahalonobis distance between two 2d arrays along the first axis.
+    Calculate the Mahalanobis distance between two 2d arrays along the first axis.
 
     Parameters
     ----------
@@ -132,7 +132,7 @@ def _mahalonobis_dist(arr1, arr2):
     Returns
     -------
     nbrs : array-like of shape (n_samples1, n_samples2)
-        Mahalonobis distance between each sample in the
+        Mahalanobis distance between each sample in the
         two input arrays.
     """
     v_inv = np.linalg.inv(np.cov(np.concatenate((arr1, arr2), axis=0).T, ddof=0))

afqinsight/nn/tests/test_pt_models.py

@@ -81,7 +81,10 @@ def run_pytorch_model(
     optimizer = torch.optim.Adam(model.parameters(), lr=0.0001)
     criterion = torch.nn.MSELoss()
     scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(
-        optimizer, mode="min", factor=0.5, patience=20, verbose=True
+        optimizer,
+        mode="min",
+        factor=0.5,
+        patience=20,
     )
 
     for epoch in range(n_epochs):

afqinsight/nn/tests/test_tf_models.py

@@ -86,7 +86,9 @@ def run_tensorflow_model(model, data_loaders, n_epochs=20):
     )
 
     reduce_lr = tf.keras.callbacks.ReduceLROnPlateau(
-        monitor="val_loss", factor=0.5, patience=20, verbose=1
+        monitor="val_loss",
+        factor=0.5,
+        patience=20,
     )
 
     callbacks = [early_stopping, ckpt, reduce_lr]

afqinsight/tests/test_match.py

@@ -34,8 +34,8 @@
     data[f"feature_{ii}"] = [*test_features[:, ii], *ctrl_features[:, ii]]
 
 
-def test_mahalonobis_dist():
-    nbrs = aim._mahalonobis_dist(test_features, ctrl_features)
+def test_mahalanobis_dist():
+    nbrs = aim._mahalanobis_dist(test_features, ctrl_features)
 
     v_inv = np.linalg.inv(
         np.cov(np.concatenate((test_features, ctrl_features), axis=0).T, ddof=0)
@@ -48,8 +48,8 @@ def test_mahalonobis_dist():
     assert_array_almost_equal(nbrs, nbrs_scipy)
 
 
-def test_mahalonobis_dist_match_df():
-    matched_df = aim.mahalonobis_dist_match(
+def test_mahalanobis_dist_match_df():
+    matched_df = aim.mahalanobis_dist_match(
         data=data,
         status_col="status",
         feature_cols=["feature_0", "feature_1", "feature_2", "feature_3"],
@@ -64,16 +64,16 @@ def test_mahalonobis_dist_match_df():
     )
 
 
-def test_mahalonobis_dist_match_df_err():
+def test_mahalanobis_dist_match_df_err():
     with pytest.raises(ValueError):  # no status column
-        aim.mahalonobis_dist_match(data=data)
+        aim.mahalanobis_dist_match(data=data)
     with pytest.raises(ValueError):  # status has more than two unique values
-        aim.mahalonobis_dist_match(data=data, status_col="feature_0")
+        aim.mahalanobis_dist_match(data=data, status_col="feature_0")
 
 
-def test_mahalonobis_dist_match_df_feauture_none():
+def test_mahalanobis_dist_match_df_feauture_none():
     data_wo_extras = data.drop(columns=["eid", "age"])
-    matched_df = aim.mahalonobis_dist_match(
+    matched_df = aim.mahalanobis_dist_match(
         data=data_wo_extras,
         status_col="status",
         threshold=1,
@@ -87,8 +87,8 @@ def test_mahalonobis_dist_match_df_feauture_none():
     )
 
 
-def test_mahalonobis_dist_match():
-    matched_df = aim.mahalonobis_dist_match(
+def test_mahalanobis_dist_match():
+    matched_df = aim.mahalanobis_dist_match(
         test=test_features, ctrl=ctrl_features, threshold=1
     )
 
@@ -103,10 +103,10 @@ def test_mahalonobis_dist_match():
     )
 
 
-def test_mahalonobis_dist_match_err():
+def test_mahalanobis_dist_match_err():
     with pytest.raises(ValueError):  # test if nan error is raised
         test_features_with_nans = test_features.copy()
         test_features_with_nans[0, 2] = np.nan
-        aim.mahalonobis_dist_match(
+        aim.mahalanobis_dist_match(
             test=test_features_with_nans, ctrl=ctrl_features, threshold=1
         )