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.

Wrong default learning_rate in SGDRegressor docstring

See original GitHub issue

Default learning_rate in SGDRegressor is ‘invscaling’, but the doc says it’s ‘optimal’. The docstring needs to be updated.

(By the way, do you known why default is ‘invscaling’ in SGDRegressor and ‘optimal’ in SGDClassifier?)

Issue Analytics

State:
Created 6 years ago
Comments:5 (5 by maintainers)

Top GitHub Comments

1reaction

amuellercommented, Oct 26, 2017

https://github.com/bodhwani/scikit-learn/blob/master/sklearn/linear_model/stochastic_gradient.py#L1259

1reaction

lestevecommented, Oct 26, 2017

@TomDLT we are trying to use “good first issue” for this kind of things rather than “Easy”. See https://github.com/scikit-learn/scikit-learn/pull/9950 for more details.

Top Results From Across the Web

1.5. Stochastic Gradient Descent - Scikit-learn

Often, an instance of SGDClassifier or SGDRegressor will have an equivalent ... For regression the default learning rate schedule is inverse scaling ......

1.5. Stochastic Gradient Descent - Scikit-learn

The class SGDRegressor implements a plain stochastic gradient descent learning routine which ... For classification, the default learning rate schedule ...

python - SGDRegressor() constantly not increasing validation ...

For best results using the default learning rate schedule, the data should have zero mean and unit variance. Details here.

scikit-learn user guide - Math-Unipd

The classes SGDClassifier and SGDRegressor provide functionality to fit linear ... For regression, the default learning rate schedule, ...

scikit-learn user guide

1.2.1 What is the project name (a lot of people get it wrong)? ... For classification, the default learning rate schedule.