You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I would expect that calling GaussianProcessRegressor.predict(X) with a single X matrix, or with repeated scalar evaluations rows of X should would give nearly the same result, but they don't.
I checked that with the RBF kernel alone, we do not have the problem.
With the DotProduct kernel alone we get:
/Users/ogrisel/code/scikit-learn/sklearn/gaussian_process/_gpr.py:629: ConvergenceWarning: lbfgs failed to converge (status=2):
ABNORMAL_TERMINATION_IN_LNSRCH.
Increase the number of iterations (max_iter) or scale the data as shown in:
https://scikit-learn.org/stable/modules/preprocessing.html
_check_optimize_result("lbfgs", opt_res)
2.0055939131680134e-08
BTW @whophil I edited your code snippet to load the Boston dataset with pandas from the original URL and be able to reproduce the problem with the latest versions of scikit-learn (including the dev branch).
Describe the bug
I would expect that calling
GaussianProcessRegressor.predict(X)
with a single X matrix, or with repeated scalar evaluations rows of X should would give nearly the same result, but they don't.An example is given below.
Steps/Code to Reproduce
Expected Results
I would expect the error to be 0 or small - perhaps around machine
eps
, if anything.Actual Results
2.1609594114124775e-08
Versions
The text was updated successfully, but these errors were encountered: