Add QuantileDMatrix support (#279)

* add quantile matrix

Signed-off-by: Zhi Lin <zhi.lin@intel.com>

* revert unrelated changes

Signed-off-by: Zhi Lin <zhi.lin@intel.com>

* add helper function

Signed-off-by: Zhi Lin <zhi.lin@intel.com>

* format

Signed-off-by: Zhi Lin <zhi.lin@intel.com>

---------

Signed-off-by: Zhi Lin <zhi.lin@intel.com>

Author: Zhi Lin

xgboost_ray/main.py

@@ -75,7 +75,7 @@ def inner_f(*args, **kwargs):
 
 from xgboost_ray.matrix import RayDMatrix, combine_data, \
     RayDeviceQuantileDMatrix, RayDataIter, concat_dataframes, \
-    LEGACY_MATRIX
+    LEGACY_MATRIX, QUANTILE_AVAILABLE, RayQuantileDMatrix
 from xgboost_ray.session import init_session, put_queue, \
     set_session_queue, get_rabit_rank
 
@@ -320,7 +320,28 @@ def _set_omp_num_threads():
     return int(float(os.environ.get("OMP_NUM_THREADS", "0.0")))
 
 
+def _prepare_dmatrix_params(param: Dict) -> Dict:
+    dm_param = {
+        "data": concat_dataframes(param["data"]),
+        "label": concat_dataframes(param["label"]),
+        "weight": concat_dataframes(param["weight"]),
+        "feature_weights": concat_dataframes(param["feature_weights"]),
+        "qid": concat_dataframes(param["qid"]),
+        "base_margin": concat_dataframes(param["base_margin"]),
+        "label_lower_bound": concat_dataframes(param["label_lower_bound"]),
+        "label_upper_bound": concat_dataframes(param["label_upper_bound"]),
+    }
+    return dm_param
+
+
 def _get_dmatrix(data: RayDMatrix, param: Dict) -> xgb.DMatrix:
+    if QUANTILE_AVAILABLE and isinstance(data, RayQuantileDMatrix):
+        if isinstance(param["data"], list):
+            qdm_param = _prepare_dmatrix_params(param)
+            param.update(qdm_param)
+        if data.enable_categorical is not None:
+            param["enable_categorical"] = data.enable_categorical
+        matrix = xgb.QuantileDMatrix(**param)
     if not LEGACY_MATRIX and isinstance(data, RayDeviceQuantileDMatrix):
         # If we only got a single data shard, create a list so we can
         # iterate over it
@@ -355,18 +376,7 @@ def _get_dmatrix(data: RayDMatrix, param: Dict) -> xgb.DMatrix:
         matrix = xgb.DeviceQuantileDMatrix(it, **dm_param)
     else:
         if isinstance(param["data"], list):
-            dm_param = {
-                "data": concat_dataframes(param["data"]),
-                "label": concat_dataframes(param["label"]),
-                "weight": concat_dataframes(param["weight"]),
-                "feature_weights": concat_dataframes(param["feature_weights"]),
-                "qid": concat_dataframes(param["qid"]),
-                "base_margin": concat_dataframes(param["base_margin"]),
-                "label_lower_bound": concat_dataframes(
-                    param["label_lower_bound"]),
-                "label_upper_bound": concat_dataframes(
-                    param["label_upper_bound"]),
-            }
+            dm_param = _prepare_dmatrix_params(param)
             param.update(dm_param)
 
         ll = param.pop("label_lower_bound", None)
@@ -669,7 +679,6 @@ def _train():
                     for deval, name in evals:
                         local_evals.append((_get_dmatrix(
                             deval, self._data[deval]), name))
-
                     if LEGACY_CALLBACK:
                         for xgb_callback in kwargs.get("callbacks", []):
                             if isinstance(xgb_callback, TrainingCallback):

xgboost_ray/matrix.py

@@ -37,6 +37,13 @@ class RayDataset:
     DataIter = object
     LEGACY_MATRIX = True
 
+try:
+    from xgboost.core import QuantileDmatrix
+    QUANTILE_AVAILABLE = True
+except ImportError:
+    QuantileDmatrix = object
+    QUANTILE_AVAILABLE = False
+
 if TYPE_CHECKING:
     from xgboost_ray.xgb import xgboost as xgb
 
@@ -875,6 +882,11 @@ def __eq__(self, other):
         return self.__hash__() == other.__hash__()
 
 
+class RayQuantileDMatrix(RayDMatrix):
+    """Currently just a thin wrapper for type detection"""
+    pass
+
+
 class RayDeviceQuantileDMatrix(RayDMatrix):
     """Currently just a thin wrapper for type detection"""