Many people think Kaggle is just a competition platform.
This could not be further from the truth. The platform has grown to include amazing learning resources no Machine Learning practitioner should miss out on.
Don't be fooled. It's a hidden goldmine, but sometimes it's hard to find things.
#1 Courses and Certificates on Kaggle Learn
Most people only know about the competitions!
Kaggle has an entire notebook-based course platform, including certificates! Whether it's SQL, Feature Engineering, or Machine Learning Explainability, you can use the integrated compute platform to learn for free. Free free, not even a computer needed!
These courses are very engaging and I particularly enjoyed both SQL courses.
#2 Find Datasets with Examples
Early steps in data science include obtaining datasets.
Kaggle has gamified uploading data sets to the platform. They have licenses and descriptions of how a dataset was obtained. Usually, a great starting point to get a free and easy data set.
Bonus: There are often code examples exploring the data already!
Just like this one!
#3 Explore Old Competitions to Learn from
I love searching old competitions for new tricks!
You don't always have to compete. Finished competitions very often have a showcase and explanation from the winners detailing how they approached the problem and how they placed in the top ranks!
More accessible than papers, free, and you can compare multiple approaches.
Here's are examples of people explaining their top solution!
#4 Find Fantastic Code When it's Hidden from You
Here's a golden nugget, trust me.
The standard sort of Kaggle Code is "Hotness", which makes sense in active competitions. When you search older code examples, click top right and switch to Most Votes to find the hidden gems from Experts to Grandmasters!
The code is currently here.
#5 Don't miss out on Discussions
Discussions may seem like a forum, sure, but it's deceptive.
Especially Grandmasters pushing for that quadruple title will often info-dump in the discussion section to farm upvotes. You can essentially get an entire paper review, outline for cross-validation, golden features, and models in a single discussion comment.
Don't skip on the discussion section!
This is one example.
Don't Miss out on an Invaluable Resource
I have yet to even get close to winning any competition.
Yet, Kaggle has benefited me in more than one way in my professional life.
And of course follow me there!