Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

sync Jun 1

Browse Source
This commit is contained in:
Your Name
2023-06-01 15:55:06 -07:00
parent c5b575e00e fa2ab1f021
commit 42867a4805
76 changed files with 1424 additions and 595 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
python-package/xgboost/core.py
View File

@@ -82,9 +82,10 @@ def from_pystr_to_cstr(data: Union[str, List[str]]) -> Union[bytes, ctypes.Array
if isinstance(data, str):
return bytes(data, "utf-8")
if isinstance(data, list):
pointers: ctypes.Array[ctypes.c_char_p] = (ctypes.c_char_p * len(data))()
data_as_bytes = [bytes(d, "utf-8") for d in data]
pointers[:] = data_as_bytes # type: ignore
data_as_bytes: List[bytes] = [bytes(d, "utf-8") for d in data]
pointers: ctypes.Array[ctypes.c_char_p] = (
ctypes.c_char_p * len(data_as_bytes)
)(*data_as_bytes)
return pointers
raise TypeError()
@@ -319,7 +320,7 @@ def _cuda_array_interface(data: DataType) -> bytes:
def ctypes2numpy(cptr: CNumericPtr, length: int, dtype: Type[np.number]) -> np.ndarray:
"""Convert a ctypes pointer array to a numpy array."""
ctype: Type[CNumeric] = _numpy2ctypes_type(dtype)
if not isinstance(cptr, ctypes.POINTER(ctype)): # type: ignore
if not isinstance(cptr, ctypes.POINTER(ctype)):
raise RuntimeError(f"expected {ctype} pointer")
res = np.zeros(length, dtype=dtype)
if not ctypes.memmove(res.ctypes.data, cptr, length * res.strides[0]):
@@ -2460,9 +2461,9 @@ class Booster:
raise TypeError("Unknown file type: ", fname)
if self.attr("best_iteration") is not None:
self.best_iteration = int(self.attr("best_iteration")) # type: ignore
self.best_iteration = int(cast(int, self.attr("best_iteration")))
if self.attr("best_score") is not None:
self.best_score = float(self.attr("best_score")) # type: ignore
self.best_score = float(cast(float, self.attr("best_score")))
def num_boosted_rounds(self) -> int:
"""Get number of boosted rounds. For gblinear this is reset to 0 after

2
python-package/xgboost/data.py
View File

@@ -882,7 +882,7 @@ def _transform_cupy_array(data: DataType) -> CupyT:
if not hasattr(data, "__cuda_array_interface__") and hasattr(data, "__array__"):
data = cupy.array(data, copy=False)
if data.dtype.hasobject or data.dtype in [cupy.float16, cupy.bool_]:
if data.dtype.hasobject or data.dtype in [cupy.bool_]:
data = data.astype(cupy.float32, copy=False)
return data

20
python-package/xgboost/spark/core.py
View File

@@ -337,11 +337,9 @@ class _SparkXGBParams(
if self.getOrDefault(self.features_cols):
if not self.getOrDefault(self.use_gpu):
raise ValueError("features_cols param requires enabling use_gpu.")
get_logger(self.__class__.__name__).warning(
"If features_cols param set, then features_col param is ignored."
)
raise ValueError(
"features_col param with list value requires enabling use_gpu."
)
if self.getOrDefault("objective") is not None:
if not isinstance(self.getOrDefault("objective"), str):
@@ -547,6 +545,8 @@ FeatureProp = namedtuple(
class _SparkXGBEstimator(Estimator, _SparkXGBParams, MLReadable, MLWritable):
_input_kwargs: Dict[str, Any]
def __init__(self) -> None:
super().__init__()
self._set_xgb_params_default()
@@ -576,6 +576,11 @@ class _SparkXGBEstimator(Estimator, _SparkXGBParams, MLReadable, MLWritable):
raise ValueError("Invalid param name: 'arbitrary_params_dict'.")
for k, v in kwargs.items():
# We're not allowing user use features_cols directly.
if k == self.features_cols.name:
raise ValueError(
f"Unsupported param '{k}' please use features_col instead."
)
if k in _inverse_pyspark_param_alias_map:
raise ValueError(
f"Please use param name {_inverse_pyspark_param_alias_map[k]} instead."
@@ -591,7 +596,10 @@ class _SparkXGBEstimator(Estimator, _SparkXGBParams, MLReadable, MLWritable):
k = real_k
if self.hasParam(k):
self._set(**{str(k): v})
if k == "features_col" and isinstance(v, list):
self._set(**{"features_cols": v})
else:
self._set(**{str(k): v})
else:
if (
k in _unsupported_xgb_params

261
python-package/xgboost/spark/estimator.py
View File

@@ -1,10 +1,13 @@
"""Xgboost pyspark integration submodule for estimator API."""
# pylint: disable=too-many-ancestors
# pylint: disable=fixme, too-many-ancestors, protected-access, no-member, invalid-name
# pylint: disable=unused-argument, too-many-locals
from typing import Any, Type
from typing import Any, Dict, List, Optional, Type, Union
import numpy as np
from pyspark import keyword_only
from pyspark.ml.param import Param, Params
from pyspark.ml.param.shared import HasProbabilityCol, HasRawPredictionCol
@@ -83,8 +86,8 @@ class SparkXGBRegressor(_SparkXGBEstimator):
:py:class:`~pyspark.ml.classification.OneVsRest`
SparkXGBRegressor automatically supports most of the parameters in
`xgboost.XGBRegressor` constructor and most of the parameters used in
:py:class:`xgboost.XGBRegressor` fit and predict method.
:py:class:`xgboost.XGBRegressor` constructor and most of the parameters used in
:py:meth:`xgboost.XGBRegressor.fit` and :py:meth:`xgboost.XGBRegressor.predict` method.
SparkXGBRegressor doesn't support setting `gpu_id` but support another param `use_gpu`,
see doc below for more details.
@@ -97,13 +100,23 @@ class SparkXGBRegressor(_SparkXGBEstimator):
SparkXGBRegressor doesn't support setting `nthread` xgboost param, instead, the `nthread`
param for each xgboost worker will be set equal to `spark.task.cpus` config value.
callbacks:
The export and import of the callback functions are at best effort.
For details, see :py:attr:`xgboost.spark.SparkXGBRegressor.callbacks` param doc.
validation_indicator_col
For params related to `xgboost.XGBRegressor` training
with evaluation dataset's supervision, set
:py:attr:`xgboost.spark.SparkXGBRegressor.validation_indicator_col`
Parameters
----------
features_col:
When the value is string, it requires the features column name to be vector type.
When the value is a list of string, it requires all the feature columns to be numeric types.
label_col:
Label column name. Default to "label".
prediction_col:
Prediction column name. Default to "prediction"
pred_contrib_col:
Contribution prediction column name.
validation_indicator_col:
For params related to `xgboost.XGBRegressor` training with
evaluation dataset's supervision,
set :py:attr:`xgboost.spark.SparkXGBRegressor.validation_indicator_col`
parameter instead of setting the `eval_set` parameter in `xgboost.XGBRegressor`
fit method.
weight_col:
@@ -111,26 +124,40 @@ class SparkXGBRegressor(_SparkXGBEstimator):
:py:attr:`xgboost.spark.SparkXGBRegressor.weight_col` parameter instead of setting
`sample_weight` and `sample_weight_eval_set` parameter in `xgboost.XGBRegressor`
fit method.
xgb_model:
Set the value to be the instance returned by
:func:`xgboost.spark.SparkXGBRegressorModel.get_booster`.
num_workers:
Integer that specifies the number of XGBoost workers to use.
Each XGBoost worker corresponds to one spark task.
use_gpu:
Boolean that specifies whether the executors are running on GPU
instances.
base_margin_col:
To specify the base margins of the training and validation
dataset, set :py:attr:`xgboost.spark.SparkXGBRegressor.base_margin_col` parameter
instead of setting `base_margin` and `base_margin_eval_set` in the
`xgboost.XGBRegressor` fit method. Note: this isn't available for distributed
training.
`xgboost.XGBRegressor` fit method.
.. Note:: The Parameters chart above contains parameters that need special handling.
For a full list of parameters, see entries with `Param(parent=...` below.
num_workers:
How many XGBoost workers to be used to train.
Each XGBoost worker corresponds to one spark task.
use_gpu:
Boolean value to specify whether the executors are running on GPU
instances.
force_repartition:
Boolean value to specify if forcing the input dataset to be repartitioned
before XGBoost training.
repartition_random_shuffle:
Boolean value to specify if randomly shuffling the dataset when repartitioning is required.
enable_sparse_data_optim:
Boolean value to specify if enabling sparse data optimization, if True,
Xgboost DMatrix object will be constructed from sparse matrix instead of
dense matrix.
kwargs:
A dictionary of xgboost parameters, please refer to
https://xgboost.readthedocs.io/en/stable/parameter.html
Note
----
The Parameters chart above contains parameters that need special handling.
For a full list of parameters, see entries with `Param(parent=...` below.
This API is experimental.
.. Note:: This API is experimental.
Examples
--------
@@ -155,9 +182,27 @@ class SparkXGBRegressor(_SparkXGBEstimator):
"""
def __init__(self, **kwargs: Any) -> None:
@keyword_only
def __init__(
self,
*,
features_col: Union[str, List[str]] = "features",
label_col: str = "label",
prediction_col: str = "prediction",
pred_contrib_col: Optional[str] = None,
validation_indicator_col: Optional[str] = None,
weight_col: Optional[str] = None,
base_margin_col: Optional[str] = None,
num_workers: int = 1,
use_gpu: bool = False,
force_repartition: bool = False,
repartition_random_shuffle: bool = False,
enable_sparse_data_optim: bool = False,
**kwargs: Dict[str, Any],
) -> None:
super().__init__()
self.setParams(**kwargs)
input_kwargs = self._input_kwargs
self.setParams(**input_kwargs)
@classmethod
def _xgb_cls(cls) -> Type[XGBRegressor]:
@@ -199,8 +244,8 @@ class SparkXGBClassifier(_SparkXGBEstimator, HasProbabilityCol, HasRawPrediction
:py:class:`~pyspark.ml.classification.OneVsRest`
SparkXGBClassifier automatically supports most of the parameters in
`xgboost.XGBClassifier` constructor and most of the parameters used in
:py:class:`xgboost.XGBClassifier` fit and predict method.
:py:class:`xgboost.XGBClassifier` constructor and most of the parameters used in
:py:meth:`xgboost.XGBClassifier.fit` and :py:meth:`xgboost.XGBClassifier.predict` method.
SparkXGBClassifier doesn't support setting `gpu_id` but support another param `use_gpu`,
see doc below for more details.
@@ -220,13 +265,21 @@ class SparkXGBClassifier(_SparkXGBEstimator, HasProbabilityCol, HasRawPrediction
Parameters
----------
callbacks:
The export and import of the callback functions are at best effort. For
details, see :py:attr:`xgboost.spark.SparkXGBClassifier.callbacks` param doc.
features_col:
When the value is string, it requires the features column name to be vector type.
When the value is a list of string, it requires all the feature columns to be numeric types.
label_col:
Label column name. Default to "label".
prediction_col:
Prediction column name. Default to "prediction"
probability_col:
Column name for predicted class conditional probabilities. Default to probabilityCol
raw_prediction_col:
The `output_margin=True` is implicitly supported by the
`rawPredictionCol` output column, which is always returned with the predicted margin
values.
pred_contrib_col:
Contribution prediction column name.
validation_indicator_col:
For params related to `xgboost.XGBClassifier` training with
evaluation dataset's supervision,
@@ -238,26 +291,39 @@ class SparkXGBClassifier(_SparkXGBEstimator, HasProbabilityCol, HasRawPrediction
:py:attr:`xgboost.spark.SparkXGBClassifier.weight_col` parameter instead of setting
`sample_weight` and `sample_weight_eval_set` parameter in `xgboost.XGBClassifier`
fit method.
xgb_model:
Set the value to be the instance returned by
:func:`xgboost.spark.SparkXGBClassifierModel.get_booster`.
num_workers:
Integer that specifies the number of XGBoost workers to use.
Each XGBoost worker corresponds to one spark task.
use_gpu:
Boolean that specifies whether the executors are running on GPU
instances.
base_margin_col:
To specify the base margins of the training and validation
dataset, set :py:attr:`xgboost.spark.SparkXGBClassifier.base_margin_col` parameter
instead of setting `base_margin` and `base_margin_eval_set` in the
`xgboost.XGBClassifier` fit method. Note: this isn't available for distributed
training.
`xgboost.XGBClassifier` fit method.
.. Note:: The Parameters chart above contains parameters that need special handling.
For a full list of parameters, see entries with `Param(parent=...` below.
num_workers:
How many XGBoost workers to be used to train.
Each XGBoost worker corresponds to one spark task.
use_gpu:
Boolean value to specify whether the executors are running on GPU
instances.
force_repartition:
Boolean value to specify if forcing the input dataset to be repartitioned
before XGBoost training.
repartition_random_shuffle:
Boolean value to specify if randomly shuffling the dataset when repartitioning is required.
enable_sparse_data_optim:
Boolean value to specify if enabling sparse data optimization, if True,
Xgboost DMatrix object will be constructed from sparse matrix instead of
dense matrix.
.. Note:: This API is experimental.
kwargs:
A dictionary of xgboost parameters, please refer to
https://xgboost.readthedocs.io/en/stable/parameter.html
Note
----
The Parameters chart above contains parameters that need special handling.
For a full list of parameters, see entries with `Param(parent=...` below.
This API is experimental.
Examples
--------
@@ -281,14 +347,34 @@ class SparkXGBClassifier(_SparkXGBEstimator, HasProbabilityCol, HasRawPrediction
"""
def __init__(self, **kwargs: Any) -> None:
@keyword_only
def __init__(
self,
*,
features_col: Union[str, List[str]] = "features",
label_col: str = "label",
prediction_col: str = "prediction",
probability_col: str = "probability",
raw_prediction_col: str = "rawPrediction",
pred_contrib_col: Optional[str] = None,
validation_indicator_col: Optional[str] = None,
weight_col: Optional[str] = None,
base_margin_col: Optional[str] = None,
num_workers: int = 1,
use_gpu: bool = False,
force_repartition: bool = False,
repartition_random_shuffle: bool = False,
enable_sparse_data_optim: bool = False,
**kwargs: Dict[str, Any],
) -> None:
super().__init__()
# The default 'objective' param value comes from sklearn `XGBClassifier` ctor,
# but in pyspark we will automatically set objective param depending on
# binary or multinomial input dataset, and we need to remove the fixed default
# param value as well to avoid causing ambiguity.
input_kwargs = self._input_kwargs
self.setParams(**input_kwargs)
self._setDefault(objective=None)
self.setParams(**kwargs)
@classmethod
def _xgb_cls(cls) -> Type[XGBClassifier]:
@@ -334,8 +420,8 @@ class SparkXGBRanker(_SparkXGBEstimator):
:py:class:`~pyspark.ml.classification.OneVsRest`
SparkXGBRanker automatically supports most of the parameters in
`xgboost.XGBRanker` constructor and most of the parameters used in
:py:class:`xgboost.XGBRanker` fit and predict method.
:py:class:`xgboost.XGBRanker` constructor and most of the parameters used in
:py:meth:`xgboost.XGBRanker.fit` and :py:meth:`xgboost.XGBRanker.predict` method.
SparkXGBRanker doesn't support setting `gpu_id` but support another param `use_gpu`,
see doc below for more details.
@@ -355,39 +441,53 @@ class SparkXGBRanker(_SparkXGBEstimator):
Parameters
----------
callbacks:
The export and import of the callback functions are at best effort. For
details, see :py:attr:`xgboost.spark.SparkXGBRanker.callbacks` param doc.
features_col:
When the value is string, it requires the features column name to be vector type.
When the value is a list of string, it requires all the feature columns to be numeric types.
label_col:
Label column name. Default to "label".
prediction_col:
Prediction column name. Default to "prediction"
pred_contrib_col:
Contribution prediction column name.
validation_indicator_col:
For params related to `xgboost.XGBRanker` training with
evaluation dataset's supervision,
set :py:attr:`xgboost.spark.XGBRanker.validation_indicator_col`
parameter instead of setting the `eval_set` parameter in `xgboost.XGBRanker`
set :py:attr:`xgboost.spark.SparkXGBRanker.validation_indicator_col`
parameter instead of setting the `eval_set` parameter in :py:class:`xgboost.XGBRanker`
fit method.
weight_col:
To specify the weight of the training and validation dataset, set
:py:attr:`xgboost.spark.SparkXGBRanker.weight_col` parameter instead of setting
`sample_weight` and `sample_weight_eval_set` parameter in `xgboost.XGBRanker`
`sample_weight` and `sample_weight_eval_set` parameter in :py:class:`xgboost.XGBRanker`
fit method.
xgb_model:
Set the value to be the instance returned by
:func:`xgboost.spark.SparkXGBRankerModel.get_booster`.
num_workers:
Integer that specifies the number of XGBoost workers to use.
Each XGBoost worker corresponds to one spark task.
use_gpu:
Boolean that specifies whether the executors are running on GPU
instances.
base_margin_col:
To specify the base margins of the training and validation
dataset, set :py:attr:`xgboost.spark.SparkXGBRanker.base_margin_col` parameter
instead of setting `base_margin` and `base_margin_eval_set` in the
`xgboost.XGBRanker` fit method.
:py:class:`xgboost.XGBRanker` fit method.
qid_col:
To specify the qid of the training and validation
dataset, set :py:attr:`xgboost.spark.SparkXGBRanker.qid_col` parameter
instead of setting `qid` / `group`, `eval_qid` / `eval_group` in the
`xgboost.XGBRanker` fit method.
Query id column name.
num_workers:
How many XGBoost workers to be used to train.
Each XGBoost worker corresponds to one spark task.
use_gpu:
Boolean value to specify whether the executors are running on GPU
instances.
force_repartition:
Boolean value to specify if forcing the input dataset to be repartitioned
before XGBoost training.
repartition_random_shuffle:
Boolean value to specify if randomly shuffling the dataset when repartitioning is required.
enable_sparse_data_optim:
Boolean value to specify if enabling sparse data optimization, if True,
Xgboost DMatrix object will be constructed from sparse matrix instead of
dense matrix.
kwargs:
A dictionary of xgboost parameters, please refer to
https://xgboost.readthedocs.io/en/stable/parameter.html
.. Note:: The Parameters chart above contains parameters that need special handling.
For a full list of parameters, see entries with `Param(parent=...` below.
@@ -426,9 +526,28 @@ class SparkXGBRanker(_SparkXGBEstimator):
>>> model.transform(df_test).show()
"""
def __init__(self, **kwargs: Any) -> None:
@keyword_only
def __init__(
self,
*,
features_col: Union[str, List[str]] = "features",
label_col: str = "label",
prediction_col: str = "prediction",
pred_contrib_col: Optional[str] = None,
validation_indicator_col: Optional[str] = None,
weight_col: Optional[str] = None,
base_margin_col: Optional[str] = None,
qid_col: Optional[str] = None,
num_workers: int = 1,
use_gpu: bool = False,
force_repartition: bool = False,
repartition_random_shuffle: bool = False,
enable_sparse_data_optim: bool = False,
**kwargs: Dict[str, Any],
) -> None:
super().__init__()
self.setParams(**kwargs)
input_kwargs = self._input_kwargs
self.setParams(**input_kwargs)
@classmethod
def _xgb_cls(cls) -> Type[XGBRanker]: