[python] fix group type in lgb.cv (#2384)

python-package/lightgbm/engine.py

@@ -307,7 +307,7 @@ def _make_n_folds(full_data, folds, nfold, params, seed, fpreproc=None, stratifi
         if hasattr(folds, 'split'):
             group_info = full_data.get_group()
             if group_info is not None:
-                group_info = group_info.astype(int)
+                group_info = np.array(group_info, dtype=int)
                 flatted_group = np.repeat(range_(len(group_info)), repeats=group_info)
             else:
                 flatted_group = np.zeros(num_data, dtype=int)
@@ -317,7 +317,7 @@ def _make_n_folds(full_data, folds, nfold, params, seed, fpreproc=None, stratifi
             if not SKLEARN_INSTALLED:
                 raise LightGBMError('Scikit-learn is required for lambdarank cv.')
             # lambdarank task, split according to groups
-            group_info = full_data.get_group().astype(int)
+            group_info = np.array(full_data.get_group(), dtype=int)
             flatted_group = np.repeat(range_(len(group_info)), repeats=group_info)
             group_kfold = _LGBMGroupKFold(n_splits=nfold)
             folds = group_kfold.split(X=np.zeros(num_data), groups=flatted_group)