[python-package] add `validate_features` argument to `refit()` (#5331)

add validate_features to refit

python-package/lightgbm/basic.py

@@ -3602,6 +3602,7 @@ class Booster:
         categorical_feature='auto',
         dataset_params=None,
         free_raw_data=True,
+        validate_features=False,
         **kwargs
     ):
         """Refit the existing Booster by new data.
@@ -3645,6 +3646,9 @@ class Booster:
             Other parameters for Dataset ``data``.
         free_raw_data : bool, optional (default=True)
             If True, raw data is freed after constructing inner Dataset for ``data``.
+        validate_features : bool, optional (default=False)
+            If True, ensure that the features used to refit the model match the original ones.
+            Used only if data is pandas DataFrame.
         **kwargs
             Other parameters for refit.
             These parameters will be passed to ``predict`` method.
@@ -3659,7 +3663,7 @@ class Booster:
         if dataset_params is None:
             dataset_params = {}
         predictor = self._to_predictor(deepcopy(kwargs))
-        leaf_preds = predictor.predict(data, -1, pred_leaf=True)
+        leaf_preds = predictor.predict(data, -1, pred_leaf=True, validate_features=validate_features)
         nrow, ncol = leaf_preds.shape
         out_is_linear = ctypes.c_int(0)
         _safe_call(_LIB.LGBM_BoosterGetLinear(

tests/python_package_test/test_engine.py

@@ -3657,3 +3657,10 @@ def test_validate_features():
 
     # check that disabling the check doesn't raise the error
     bst.predict(df2, validate_features=False)
+
+    # try to refit with a different feature
+    with pytest.raises(lgb.basic.LightGBMError, match="Expected 'x3' at position 2 but found 'z'"):
+        bst.refit(df2, y, validate_features=True)
+
+    # check that disabling the check doesn't raise the error
+    bst.refit(df2, y, validate_features=False)