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

Enable categorical data support on Python DMatrix. (#6166)

Browse Source 
* Only pandas is recognized.
This commit is contained in:
Jiaming Yuan 2020-09-29 11:22:56 +08:00 committed by GitHub
parent 52c0b3f100
commit 7622b8cdb8
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 57 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
python-package/xgboost/core.py
View File

@ -384,7 +384,8 @@ class DMatrix: # pylint: disable=too-many-instance-attributes
silent=False, silent=False,
feature_names=None, feature_names=None,
feature_types=None, feature_types=None,
nthread=None): nthread=None,
enable_categorical=False):
"""Parameters """Parameters
---------- ----------
data : os.PathLike/string/numpy.array/scipy.sparse/pd.DataFrame/ data : os.PathLike/string/numpy.array/scipy.sparse/pd.DataFrame/
@ -419,6 +420,16 @@ class DMatrix: # pylint: disable=too-many-instance-attributes
Number of threads to use for loading data when parallelization is Number of threads to use for loading data when parallelization is
applicable. If -1, uses maximum threads available on the system. applicable. If -1, uses maximum threads available on the system.
enable_categorical: boolean, optional
.. versionadded:: 1.3.0
Experimental support of specializing for categorical features. Do
not set to True unless you are interested in development.
Currently it's only available for `gpu_hist` tree method with 1 vs
rest (one hot) categorical split. Also, JSON serialization format,
`gpu_predictor` and pandas input are required.
""" """
if isinstance(data, list): if isinstance(data, list):
raise TypeError('Input data can not be a list.') raise TypeError('Input data can not be a list.')
@ -437,7 +448,8 @@ class DMatrix: # pylint: disable=too-many-instance-attributes
data, missing=self.missing, data, missing=self.missing,
threads=self.nthread, threads=self.nthread,
feature_names=feature_names, feature_names=feature_names,
feature_types=feature_types) feature_types=feature_types,
enable_categorical=enable_categorical)
assert handle is not None assert handle is not None
self.handle = handle self.handle = handle

36
python-package/xgboost/data.py
View File

@ -184,20 +184,24 @@ _pandas_dtype_mapper = {
} }
def _transform_pandas_df(data, feature_names=None, feature_types=None, def _transform_pandas_df(data, enable_categorical,
feature_names=None, feature_types=None,
meta=None, meta_type=None): meta=None, meta_type=None):
from pandas import MultiIndex, Int64Index from pandas import MultiIndex, Int64Index
from pandas.api.types import is_sparse from pandas.api.types import is_sparse, is_categorical
data_dtypes = data.dtypes data_dtypes = data.dtypes
if not all(dtype.name in _pandas_dtype_mapper or is_sparse(dtype) if not all(dtype.name in _pandas_dtype_mapper or is_sparse(dtype) or
(is_categorical(dtype) and enable_categorical)
for dtype in data_dtypes): for dtype in data_dtypes):
bad_fields = [ bad_fields = [
str(data.columns[i]) for i, dtype in enumerate(data_dtypes) str(data.columns[i]) for i, dtype in enumerate(data_dtypes)
if dtype.name not in _pandas_dtype_mapper if dtype.name not in _pandas_dtype_mapper
] ]
msg = """DataFrame.dtypes for data must be int, float or bool. msg = """DataFrame.dtypes for data must be int, float, bool or categorical. When
Did not expect the data types in fields """ categorical type is supplied, DMatrix parameter
`enable_categorical` must be set to `True`."""
raise ValueError(msg + ', '.join(bad_fields)) raise ValueError(msg + ', '.join(bad_fields))
if feature_names is None and meta is None: if feature_names is None and meta is None:
@ -216,6 +220,8 @@ def _transform_pandas_df(data, feature_names=None, feature_types=None,
if is_sparse(dtype): if is_sparse(dtype):
feature_types.append(_pandas_dtype_mapper[ feature_types.append(_pandas_dtype_mapper[
dtype.subtype.name]) dtype.subtype.name])
elif is_categorical(dtype) and enable_categorical:
feature_types.append('categorical')
else: else:
feature_types.append(_pandas_dtype_mapper[dtype.name]) feature_types.append(_pandas_dtype_mapper[dtype.name])
@ -226,13 +232,13 @@ def _transform_pandas_df(data, feature_names=None, feature_types=None,
dtype = meta_type if meta_type else np.float32 dtype = meta_type if meta_type else np.float32
data = np.ascontiguousarray(data.values, dtype=dtype) data = np.ascontiguousarray(data.values, dtype=dtype)
return data, feature_names, feature_types return data, feature_names, feature_types
def _from_pandas_df(data, missing, nthread, feature_names, feature_types): def _from_pandas_df(data, enable_categorical, missing, nthread,
feature_names, feature_types):
data, feature_names, feature_types = _transform_pandas_df( data, feature_names, feature_types = _transform_pandas_df(
data, feature_names, feature_types) data, enable_categorical, feature_names, feature_types)
return _from_numpy_array(data, missing, nthread, feature_names, return _from_numpy_array(data, missing, nthread, feature_names,
feature_types) feature_types)
@ -244,6 +250,7 @@ def _is_pandas_series(data):
return False return False
return isinstance(data, pd.Series) return isinstance(data, pd.Series)
def _is_modin_series(data): def _is_modin_series(data):
try: try:
import modin.pandas as pd import modin.pandas as pd
@ -507,7 +514,8 @@ def _has_array_protocol(data):
def dispatch_data_backend(data, missing, threads, def dispatch_data_backend(data, missing, threads,
feature_names, feature_types): feature_names, feature_types,
enable_categorical=False):
'''Dispatch data for DMatrix.''' '''Dispatch data for DMatrix.'''
if _is_scipy_csr(data): if _is_scipy_csr(data):
return _from_scipy_csr(data, missing, feature_names, feature_types) return _from_scipy_csr(data, missing, feature_names, feature_types)
@ -525,7 +533,7 @@ def dispatch_data_backend(data, missing, threads,
if _is_tuple(data): if _is_tuple(data):
return _from_tuple(data, missing, feature_names, feature_types) return _from_tuple(data, missing, feature_names, feature_types)
if _is_pandas_df(data): if _is_pandas_df(data):
return _from_pandas_df(data, missing, threads, return _from_pandas_df(data, enable_categorical, missing, threads,
feature_names, feature_types) feature_names, feature_types)
if _is_pandas_series(data): if _is_pandas_series(data):
return _from_pandas_series(data, missing, threads, feature_names, return _from_pandas_series(data, missing, threads, feature_names,
@ -551,7 +559,7 @@ def dispatch_data_backend(data, missing, threads,
return _from_dt_df(data, missing, threads, feature_names, return _from_dt_df(data, missing, threads, feature_names,
feature_types) feature_types)
if _is_modin_df(data): if _is_modin_df(data):
return _from_pandas_df(data, missing, threads, return _from_pandas_df(data, enable_categorical, missing, threads,
feature_names, feature_types) feature_names, feature_types)
if _is_modin_series(data): if _is_modin_series(data):
return _from_pandas_series(data, missing, threads, feature_names, return _from_pandas_series(data, missing, threads, feature_names,
@ -655,7 +663,8 @@ def dispatch_meta_backend(matrix: DMatrix, data, name: str, dtype: str = None):
_meta_from_numpy(data, name, dtype, handle) _meta_from_numpy(data, name, dtype, handle)
return return
if _is_pandas_df(data): if _is_pandas_df(data):
data, _, _ = _transform_pandas_df(data, meta=name, meta_type=dtype) data, _, _ = _transform_pandas_df(data, False, meta=name,
meta_type=dtype)
_meta_from_numpy(data, name, dtype, handle) _meta_from_numpy(data, name, dtype, handle)
return return
if _is_pandas_series(data): if _is_pandas_series(data):
@ -680,7 +689,8 @@ def dispatch_meta_backend(matrix: DMatrix, data, name: str, dtype: str = None):
_meta_from_dt(data, name, dtype, handle) _meta_from_dt(data, name, dtype, handle)
return return
if _is_modin_df(data): if _is_modin_df(data):
data, _, _ = _transform_pandas_df(data, meta=name, meta_type=dtype) data, _, _ = _transform_pandas_df(
data, False, meta=name, meta_type=dtype)
_meta_from_numpy(data, name, dtype, handle) _meta_from_numpy(data, name, dtype, handle)
return return
if _is_modin_series(data): if _is_modin_series(data):

9
tests/python/test_with_modin.py
View File

@ -67,7 +67,8 @@ class TestModin(unittest.TestCase):
# 0 1 1 0 0 # 0 1 1 0 0
# 1 2 0 1 0 # 1 2 0 1 0
# 2 3 0 0 1 # 2 3 0 0 1
result, _, _ = xgb.data._transform_pandas_df(dummies) result, _, _ = xgb.data._transform_pandas_df(dummies,
enable_categorical=False)
exp = np.array([[1., 1., 0., 0.], exp = np.array([[1., 1., 0., 0.],
[2., 0., 1., 0.], [2., 0., 1., 0.],
[3., 0., 0., 1.]]) [3., 0., 0., 1.]])
@ -113,15 +114,15 @@ class TestModin(unittest.TestCase):
# label must be a single column # label must be a single column
df = md.DataFrame({'A': ['X', 'Y', 'Z'], 'B': [1, 2, 3]}) df = md.DataFrame({'A': ['X', 'Y', 'Z'], 'B': [1, 2, 3]})
self.assertRaises(ValueError, xgb.data._transform_pandas_df, df, self.assertRaises(ValueError, xgb.data._transform_pandas_df, df,
None, None, 'label', 'float') False, None, None, 'label', 'float')
# label must be supported dtype # label must be supported dtype
df = md.DataFrame({'A': np.array(['a', 'b', 'c'], dtype=object)}) df = md.DataFrame({'A': np.array(['a', 'b', 'c'], dtype=object)})
self.assertRaises(ValueError, xgb.data._transform_pandas_df, df, self.assertRaises(ValueError, xgb.data._transform_pandas_df, df,
None, None, 'label', 'float') False, None, None, 'label', 'float')
df = md.DataFrame({'A': np.array([1, 2, 3], dtype=int)}) df = md.DataFrame({'A': np.array([1, 2, 3], dtype=int)})
result, _, _ = xgb.data._transform_pandas_df(df, None, None, result, _, _ = xgb.data._transform_pandas_df(df, False, None, None,
'label', 'float') 'label', 'float')
np.testing.assert_array_equal(result, np.array([[1.], [2.], [3.]], np.testing.assert_array_equal(result, np.array([[1.], [2.], [3.]],
dtype=float)) dtype=float))

19
tests/python/test_with_pandas.py
View File

@ -67,7 +67,8 @@ class TestPandas(unittest.TestCase):
# 0 1 1 0 0 # 0 1 1 0 0
# 1 2 0 1 0 # 1 2 0 1 0
# 2 3 0 0 1 # 2 3 0 0 1
result, _, _ = xgb.data._transform_pandas_df(dummies) result, _, _ = xgb.data._transform_pandas_df(dummies,
enable_categorical=False)
exp = np.array([[1., 1., 0., 0.], exp = np.array([[1., 1., 0., 0.],
[2., 0., 1., 0.], [2., 0., 1., 0.],
[3., 0., 0., 1.]]) [3., 0., 0., 1.]])
@ -109,6 +110,16 @@ class TestPandas(unittest.TestCase):
assert dm.num_row() == 2 assert dm.num_row() == 2
assert dm.num_col() == 6 assert dm.num_col() == 6
def test_pandas_categorical(self):
rng = np.random.RandomState(1994)
rows = 100
X = rng.randint(3, 7, size=rows)
X = pd.Series(X, dtype="category")
X = pd.DataFrame({'f0': X})
y = rng.randn(rows)
m = xgb.DMatrix(X, y, enable_categorical=True)
assert m.feature_types[0] == 'categorical'
def test_pandas_sparse(self): def test_pandas_sparse(self):
import pandas as pd import pandas as pd
rows = 100 rows = 100
@ -129,15 +140,15 @@ class TestPandas(unittest.TestCase):
# label must be a single column # label must be a single column
df = pd.DataFrame({'A': ['X', 'Y', 'Z'], 'B': [1, 2, 3]}) df = pd.DataFrame({'A': ['X', 'Y', 'Z'], 'B': [1, 2, 3]})
self.assertRaises(ValueError, xgb.data._transform_pandas_df, df, self.assertRaises(ValueError, xgb.data._transform_pandas_df, df,
None, None, 'label', 'float') False, None, None, 'label', 'float')
# label must be supported dtype # label must be supported dtype
df = pd.DataFrame({'A': np.array(['a', 'b', 'c'], dtype=object)}) df = pd.DataFrame({'A': np.array(['a', 'b', 'c'], dtype=object)})
self.assertRaises(ValueError, xgb.data._transform_pandas_df, df, self.assertRaises(ValueError, xgb.data._transform_pandas_df, df,
None, None, 'label', 'float') False, None, None, 'label', 'float')
df = pd.DataFrame({'A': np.array([1, 2, 3], dtype=int)}) df = pd.DataFrame({'A': np.array([1, 2, 3], dtype=int)})
result, _, _ = xgb.data._transform_pandas_df(df, None, None, result, _, _ = xgb.data._transform_pandas_df(df, False, None, None,
'label', 'float') 'label', 'float')
np.testing.assert_array_equal(result, np.array([[1.], [2.], [3.]], np.testing.assert_array_equal(result, np.array([[1.], [2.], [3.]],
dtype=float)) dtype=float))