Editorial Summary :

Linear regression is among the primary/entry-level Machine Learning (ML) models . It’s possible to endow the regression coefficients with some extra properties by considering the MSE plus an additional penalty term . With a particular choice of the penalty term, we can provide linear regression with a feature selection property . We can use Group Lasso to perform feature selection of a mixture of continuous/categorical variables . Continuous variables will be considered as groups with one variable and categorical ones will be treated as groups of dummies . celer is a python package that embraces the scikit-learn vision and provides models that are entirely designed under scikits-learn API and thereby integrate well with it . Celer is specially designed to handle Lasso-like models such as Lasso and Group Lasso . It features a tailored implementation that enables it to fit these types of models quickly . celer can be used with Pipelines, GridSearchCV among others . After installing celer via pip, you can easily fit a Group . Lasso model as shown below . Linear regression remains one of the simplest and easy-to-understand models which coupled with well-chosen penalties yields more interpretable results . In this article, we extended linear regression to perform feature selection of a mixture of continuous/categorical variables using the Group Lasso model . In the end, we used celer to fit the model as well as relied on its core solver — celer_path — to control the number of selected features . You can check the source code in this link, and leverage the developed python utils to apply it to your own use case .

Key Highlights :

  • Linear regression is among the primary/entry-level Machine Learning (ML) models .
  • It owes that to its… It owes . that to the “Hello world” program for Data scientists .
  • celer is a python package that embraces the scikit-learn vision and provides models that are entirely designed under scikits-learn API .
  • Linear regression remains one of the simplest and easy-to-understand models .
  • In this article, we extended linear regression to perform feature selection of a mixture of continuous/categorical variables using the Group Lasso model .

The editorial is based on the content sourced from towardsdatascience.com

Read the full article.

Similar Posts