Cross validation error of a Gaussian process with noisy target #26328
Description
Describe the bug
Hi,
I'm trying to use RandomizedSearchCV with GP with a vector of alpha's.
It seems that with the cross validation the alpha are not being split into train/test sets because I get the following error:
ValueError: alpha must be a scalar or an array with same number of entries as y. (100 != 80)
Is there any workaround, or am I missing something?
Thanks!
Steps/Code to Reproduce
import numpy as np
from sklearn.gaussian_process import GaussianProcessRegressor
from sklearn.gaussian_process.kernels import RBF
from sklearn.model_selection import RandomizedSearchCV
X = np.random.rand(100, 2)
y = np.sin(X[:, 0]) + np.cos(X[:, 1])
alpha = 0.1 * np.ones_like(y)
kernel = RBF(length_scale=1.)
param_dist = {
'kernel__length_scale': (1e-3, 1e3),
}
gp = GaussianProcessRegressor(kernel=kernel, alpha=alpha)
rs = RandomizedSearchCV(gp, param_distributions=param_dist, cv=5, n_iter=50)
rs.fit(X, y)
print("Best hyperparameters: ", rs.best_params_)
print("Best score: ", rs.best_score_)Expected Results
No error is thrown and estimation is based on the values and target uncertainty
Actual Results
ValueError: alpha must be a scalar or an array with same number of entries as y. (100 != 80)Versions
System:
python: 3.10.10 | packaged by conda-forge | (main, Mar 24 2023, 20:12:31) [Clang 14.0.6 ]
machine: macOS-13.3.1-arm64-arm-64bit
Python dependencies:
sklearn: 1.2.2
pip: 23.1.2
setuptools: 67.7.2
numpy: 1.23.5
scipy: 1.10.0
Cython: None
pandas: 1.5.2
matplotlib: 3.6.2
joblib: 1.2.0
threadpoolctl: 3.1.0
Built with OpenMP: True