GradientBoostingRegressor initial estimator does not play together with Pipeline #13466
Description
Using a pipeline as the initial estimator of GradientBoostingRegressor doesn't work due to incompatible signatures.
import sklearn
import sklearn.pipeline
import sklearn.ensemble
import sklearn.decomposition
import sklearn.linear_model
import numpy as np
init = sklearn.pipeline.make_pipeline(sklearn.decomposition.PCA(), sklearn.linear_model.ElasticNet())
model = sklearn.ensemble.GradientBoostingRegressor(init=init)
x = np.random.rand(12, 3)
y = np.random.rand(12)
model.fit(x, y)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Users/Thomas/.local/miniconda3/envs/4cast/lib/python3.6/site-packages/sklearn/ensemble/gradient_boosting.py", line 1421, in fit
self.init_.fit(X, y, sample_weight)
TypeError: fit() takes from 2 to 3 positional arguments but 4 were givenThe signature of Pipeline.fit is
# sklearn/pipeline.py
...
239 def fit(self, X, y=None, **fit_params):
...which cannot be called with three positional arguments as above.
So I guess line 1421 in sklearn/ensemble/gradient_boosting.py should read
self.init_.fit(X, y, sample_weight=sample_weight) instead and the issue is solved. Right?
Versions
>>> sklearn.show_versions()
System:
python: 3.6.2 |Continuum Analytics, Inc.| (default, Jul 20 2017, 13:14:59) [GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.57)]
executable: /Users/Thomas/.local/miniconda3/envs/test/bin/python
machine: Darwin-18.2.0-x86_64-i386-64bit
BLAS:
macros: SCIPY_MKL_H=None, HAVE_CBLAS=None
lib_dirs: /Users/Thomas/.local/miniconda3/envs/test/lib
cblas_libs: mkl_rt, pthread
Python deps:
pip: 10.0.1
setuptools: 39.2.0
sklearn: 0.20.2
numpy: 1.16.1
scipy: 1.2.0
Cython: None
pandas: 0.24.2