Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

scikit-learn's Gaussian Process bug might be affecting your results

See original GitHub issue

This is not necessarily an issue with bayes_opt but it’s highly related and therefore I thought I should let you know about it.

Recently, I found a bug that I recently found in scikit-learn’s Gaussian Process module that might be affecting the results you get from bayes_opt since bayes_opt uses scikit-learn’s GP module as the surrogate model. However, in case bayes_opt only depends on the relative standard deviation values, which I think is the case, you will be safe and probably the bug won’t be affecting you that much. That’s why I asked the scikit-learn developers in one of the comments to add a flag whose False value will basically return the relative std values that are within [0, 1]. But in case bayes_opt depends on the actual values of the standard deviation it is almost certain that you need to run your experiments again once this bug is fixed.

Also, the scikit-learn developers say they are busy at the moment and who knows when they’re gonna be able to fix the issue. So maybe you can help them fix the bug and fix this issue as soon as possible.

Issue Analytics

State:
Created 4 years ago
Reactions:1
Comments:6 (2 by maintainers)

Top GitHub Comments

1reaction

Amir-Arsalancommented, Jan 21, 2020

The scikit-learn developers are fixing this issue. Here’s a pull request for it.

0reactions

ben-arnaocommented, Jul 14, 2020

@Amir-Arsalan This bug has been fixed. Would you mind closing the issue as to not confuse anyone? Thank you.

Top Results From Across the Web

sklearn.gaussian_process.GaussianProcessRegressor

Draw samples from Gaussian process and evaluate at X. score (X, y[, sample_weight]). Return the coefficient of determination of the prediction.

Multi-output spatial statistics with gaussian processes

First off you need to split your data. Training a model and then predicting on that same training data will look like overfitting...

Understanding Gaussian Process, the Socratic Way | by Wei Yi

Being a Bayesian method, Gaussian Process makes predictions with uncertainty. For example, it will predict that tomorrow's stock price is $100, ...

SAMPL6 Challenge Results from pKa Predictions Based on a ...

These features are fed into a Scikit-learn Gaussian process to predict ... estimate based on the chemistry of the molecule that would reflect...

scikit-learn 0.16.1 documentation

Various fixes to the Gaussian processes subpackage by Vincent Dubourg and Jan ... This bug fix will affect computational and generalization performance of ......