I’ll be bold (again): Feature-engine is the go to library for feature selection
Why Feature-engine Should Be Your Top Choice for Feature Selection
Feature-engine supports the following feature selection methods:
Selection based on feature cardinality (which is a measure of variability also applicable to categorical variables)
Removes duplicate features
Removes correlated features (brute force or smartly)
Selects features based on information gain
Selects features based on Population Stability Index (PSI)
Selects features based on permutation feature importance
Selects features by comparing them with random variables (called probes)
Recursive feature elimination evaluating model performance
Rrecursive feature addition evaluating model performance
Selects features by target mean encoding (super fast and allows you to select among categorical variables prior to encoding them)
Selects features based on single feature classifiers or regressors
MRMR
Let’s now take a look at scikit-learn’s feature selection methods:
Removes features with low variance
Selects features by tree-derived importance (all in one go, or through recursive elimination)
Selects features based on lasso regularization (all in one go, or through recursive elimination)
Selects features based on anova or correlation against the target
Selects features with high mutual information with the target
Sequential feature elimination (wrapper method)
Sequential feature addition (wrapper method)
And there is a third library, MLXtend, which supports:
Sequential feature elimination (wrapper method)
Sequential feature addition (wrapper method)
Don’t get me wrong—this isn’t a competition. These tools democratize well-known methods for uncovering patterns in our data and relationships among variables.
I want to highlight that if you haven’t explored Feature-engine for feature selection, you might be missing out on valuable algorithms that aren’t supported elsewhere.
So, what are you waiting for? Check it out: 👇
https://feature-engine.trainindata.com/en/latest/user_guide/selection/index.html
To learn more about ways to assign feature importance for selection check out:
That's not all—we have an exciting announcement! 🎉
We've been working behind the scenes to enhance your experience, and we're thrilled to unveil our freshly revamped website! 🚀
Our Website Just Got a Refresh!
We’ve given our website a new look and added some great features to help you find what you are looking for:
✅ New Specializations Page: Check out our new Specializations page—showing our bundles of focused courses designed to boost your skills in specific domains. (add link to page)
✅ New Team Page: Meet the creators behind the screens, the experts who make it all happen.
✅ New Newsletters Page: In case you didn't know, we have a newsletter. Why should you subscribe? Judge it yourself: here you can check out some of our latest editions and discover if this is something you don’t want to miss.
We’re always working to make your learning experience better. Take a look and let us know what you think!
🖥 To our website
Ready to enhance your skills?
Our specializations, courses and books are here to assist you:
Advanced Machine Learning (specialization)
Forecasting with Machine Learning (course)