Merge pull request #103 from nyanp/bugfix/fix-ci-errors

Fix ci errors, close #101, #102

nyaggle/feature/category_encoder/target_encoder.py

@@ -174,7 +174,7 @@ class TargetEncoder(KFoldEncoderWrapper):
                  groups: Optional[pd.Series] = None,
                  cols: List[str] = None,
                  drop_invariant: bool = False, handle_missing: str = 'value', handle_unknown: str = 'value',
-                 min_samples_leaf: int = 1, smoothing: float = 1.0, return_same_type: bool = True):
+                 min_samples_leaf: int = 20, smoothing: float = 10.0, return_same_type: bool = True):
         e = ce.TargetEncoder(cols=cols, drop_invariant=drop_invariant, return_df=True,
                              handle_missing=handle_missing,
                              handle_unknown=handle_unknown,

nyaggle/feature/nlp/bert.py

@@ -116,7 +116,7 @@ class BertSentenceVectorizer(BaseFeaturizer):
         X = convert_input(X)
 
         tqdm.pandas()
-        columns = self.text_columns or [c for c in X.columns if X[c].dtype == np.object]
+        columns = self.text_columns or [c for c in X.columns if X[c].dtype == object]
         non_text_columns = [c for c in X.columns if c not in columns]
 
         column_names = []

nyaggle/validation/split.py

@@ -450,7 +450,7 @@ class StratifiedGroupKFold(_BaseKFold):
             split_diffs[splitx] = cand_diffs[splitx]
             grouped_splitx.append(splitx)
 
-        test_folds = np.empty(n_samples, dtype=np.int)
+        test_folds = np.empty(n_samples, dtype=int)
         for group_idx, splitx in zip(sortx, grouped_splitx):
             idxs = group_idxs[group_idx]
             test_folds[idxs] = splitx