[c-api][python-package][R-package] expose feature num bin

include/LightGBM/c_api.h

@@ -432,6 +432,17 @@ LIGHTGBM_C_EXPORT int LGBM_DatasetGetNumData(DatasetHandle handle,
 LIGHTGBM_C_EXPORT int LGBM_DatasetGetNumFeature(DatasetHandle handle,
                                                 int* out);
 
+/*!
+ * \brief Get number of bins for feature.
+ * \param handle Handle of dataset
+ * \param feature Index of the feature
+ * \param[out] out The address to hold number of bins
+ * \return 0 when succeed, -1 when failure happens
+ */
+LIGHTGBM_C_EXPORT int LGBM_DatasetGetFeatureNumBin(DatasetHandle handle,
+                                                   int feature,
+                                                   int* out);
+
 /*!
  * \brief Add features from ``source`` to ``target``.
  * \param target The handle of the dataset to add features to

python-package/lightgbm/basic.py

@@ -2371,6 +2371,29 @@ def num_feature(self):
         else:
             raise LightGBMError("Cannot get num_feature before construct dataset")
 
+    def feature_num_bin(self, feature_idx: int) -> int:
+        """Get the number of bins for a feature.
+
+        Parameters
+        ----------
+        feature_idx : int
+            Index of the feature.
+
+        Returns
+        -------
+        number_of_bins : int
+            The number of constructed bins for the feature in the Dataset.
+        """
+        if self.handle is not None:
+            ret = ctypes.c_int(0)
+            feature = ctypes.c_int(feature_idx)
+            _safe_call(_LIB.LGBM_DatasetGetFeatureNumBin(self.handle,
+                                                         feature,
+                                                         ctypes.byref(ret)))
+            return ret.value
+        else:
+            raise LightGBMError("Cannot get feature_num_bin before construct dataset")
+
     def get_ref_chain(self, ref_limit=100):
         """Get a chain of Dataset objects.
 

src/c_api.cpp

@@ -1550,6 +1550,20 @@ int LGBM_DatasetGetNumFeature(DatasetHandle handle,
   API_END();
 }
 
+int LGBM_DatasetGetFeatureNumBin(DatasetHandle handle,
+                                 int feature,
+                                 int* out) {
+  API_BEGIN();
+  auto dataset = reinterpret_cast<Dataset*>(handle);
+  int inner_idx = dataset->InnerFeatureIndex(feature);
+  if (inner_idx >= 0) {
+    *out = dataset->FeatureNumBin(inner_idx);
+  } else {
+    *out = 0;
+  }
+  API_END();
+}
+
 int LGBM_DatasetAddFeaturesFrom(DatasetHandle target,
                                 DatasetHandle source) {
   API_BEGIN();

tests/python_package_test/test_basic.py

@@ -621,3 +621,23 @@ def test_no_copy_when_single_float_dtype_dataframe(dtype):
     built_data = lgb.basic._data_from_pandas(df, feature_name, None, None)[0]
     assert built_data.dtype == dtype
     assert np.shares_memory(X, built_data)
+
+
+def test_feature_num_bin():
+    X = np.vstack([
+        np.random.rand(100),
+        np.array([1, 2] * 50),
+        np.array([0, 1, 2] * 33 + [0]),
+        np.array([1, 2] * 49 + 2 * [np.nan]),
+        np.zeros(100),
+    ]).T
+    ds = lgb.Dataset(X).construct()
+    expected_num_bins = np.array([
+        35,  # ceil(100[n_samples] / 3[min_data_in_bin]) = 34 + bin for zero
+        3,  # 0, 1, 2
+        3,  # 0, 1, 2
+        4,  # 0, 1, 2 + nan
+        0,  # unused
+    ])
+    actual_num_bins = [ds.feature_num_bin(i) for i in range(X.shape[1])]
+    np.testing.assert_equal(actual_num_bins, expected_num_bins)