Kaggle Learn

What is on offer

Kaggle Learn is a set of micro-courses, each a handful of short lessons followed by an exercise notebook you complete in the browser. Kaggle describes each course as taking between one and seven hours. Topics include:

• Foundations: Python, pandas, data visualisation, data cleaning, Intro to SQL and Advanced SQL.
• Machine learning: Intro to Machine Learning, Intermediate Machine Learning (missing values, categorical data, pipelines, data leakage), Feature Engineering and Machine Learning Explainability.
• Deep learning and beyond: Intro to Deep Learning (with TensorFlow and Keras), Computer Vision, Time Series and Geospatial Analysis, plus shorter courses on topics such as AI ethics.

Courses are free and each gives a completion certificate.

A good order

For someone new to data work:

1. Python (skip if you already program in Python).
2. Pandas and Data Visualization.
3. Intro to Machine Learning, which ends with a house price competition entry.
4. Intermediate Machine Learning, where data leakage and pipelines are covered; this is the most valuable course for avoiding misleading results.
5. Feature Engineering, then Intro to SQL if you are heading towards data analyst or backend roles.

At a few hours each, you can realistically finish the first four in two weekends.

Why the short format works

The barrier to starting machine learning is usually setup and scope, not difficulty. A course you finish today builds momentum; a forty-hour course abandoned in week two does not. Because everything runs in Kaggle notebooks, there is no environment to install, and you practise on real datasets from the first lesson.

Where it stops

These are introductions. They teach you how to use the tools, but not much theory, and they will not make you job-ready alone. The certificates are a pleasant record of progress, but carry little weight with employers. After a few courses, move on to:

• A structured course with more theory, such as Google's Machine Learning Crash Course or the scikit-learn MOOC.
• Real practice in Kaggle's Getting Started competitions.
• A small project of your own on data you care about.

Common questions

Do I need to install anything? No. You need a free Kaggle account and a browser.

Are the certificates worth listing? You can list them, but a project or notebook you built will say much more.

Is it good for SQL? The two SQL courses use Google BigQuery public datasets and are a solid, quick introduction.

Best for: complete beginners, and engineers who want a quick working knowledge of pandas, SQL or feature engineering.