Your learning journey matters and we’d love to hear it

If you’re currently taking our Forecasting with Machine Learning or Feature Engineering for Time Series Forecasting courses, we wanted to say thank you 🙏

We’d also love to invite you to share your experience on Linkedin. 

Building these courses has been a labour of love for Kishan and me. Seeing how learners apply forecasting concepts in their own work is what truly motivates us.

Just share whatever feels genuine to you — a few honest lines really mean a lot to us and help others discover forecasting with confidence.

Please tag both Kishan and me so we can see and interact with your your post. As a thank you, you'll get 30% off your next course, book, or specialisation. Once your LinkedIn post is live, I’ll DM you a discount code.

Thanks for being part of this learning journey — we’re truly grateful to have you here.

Feature Selection: 5 Python Libraries Worth Knowing

Feature selection doesn’t always get the spotlight but it probably should.

If you’ve ever trained a model and thought, “Why is this so slow?” or “Which of these features actually matter?”...feature selection is often the answer.

Here are a few Python libraries that make feature selection a lot more approachable (and sometimes even fun):

🔍 scikit-feature
Think of this as a feature selection playground.
It’s a large repository that brings together many different feature selection methods in one place. It's great if you like exploring and comparing approaches.

🌳 boruta_py
If your goal is to find all features that truly matter, not just the bare minimum, Boruta is a great option. It works nicely with scikit-learn and is especially popular with tree-based models.

⚡BoostARoota
This one’s built on XGBoost and focuses on speed. It’s a practical choice when you’re working with larger datasets and want a fast, automated way to trim down your features.

🧠 scikit-rebate
Relief-based methods are really good at uncovering feature interactions. The kind that simple correlation checks miss. scikit-rebate wraps these ideas into a scikit-learn-friendly package.

🧬 zoofs
Zoofs takes a different route by using evolutionary algorithms to search for strong feature subsets.It’s especially useful when the feature space is complex and you don’t want to rely on greedy selection rules.

Why bother with feature selection? Because fewer, better features usually mean:
👉🏻 Cleaner models
👉🏻 Better generalisation
👉🏻 Faster training
👉🏻 Easier explanations

And honestly debugging models becomes much less painful.

I hope this information was useful!



Wishing you a successful week ahead - see you next Monday! 👋🏻

Sole

While you are at it, check out our courses:

Stop aimless internet browsing. Start learning today with meticulously crafted courses offering a robust curriculum, fostering skill development with steadfast focus and efficiency.

• Forecasting specialization (course)
• Interpreting Machine Learning Models (course)
• Python Feature Engineering Cookbook (book)

Did someone share this email with you? Think it's pretty cool? Then just hit the button and subscribe to Data Bites. Don’t miss out on any of our tips and propel your data science career to new heights.