The classic curriculum: regression, classification, clustering, decision trees, ensembles, and the fundamentals every AI practitioner is expected to know. Includes legendary free university lectures and structured paid programs.
Sorted cheapest first. Click a column header to re-sort. Aspiring data scientists and ML engineers.
| Course | Level | Duration | Price | Certificate | |
|---|---|---|---|---|---|
Kaggle |
beginner | ~3 hours (7 lessons) | Free | Yes | Visit |
MIT · MIT OpenCourseWare |
intermediate | 13 weeks (one semester) | Free | No | Visit |
Caltech |
intermediate | 18 lectures plus homework | Free | No | Visit |
Google |
beginner | ~15 hours (12 modules) | Free | No | Visit |
Microsoft · GitHub |
beginner | 12 weeks, 26 lessons | Free | No | Visit |
Stanford University · YouTube |
advanced | 20 lectures, ~1.5 hours each | Free | No | Visit |
Stanford Online / DeepLearning.AI · Coursera |
beginner | 2 months at 10 hrs/week (~95 hours, 3 courses) | Free to audit Free to audit · $49/mo |
Yes | Visit |
Stanford University · Stanford Online |
advanced | 10 weeks at 10-15 hrs/week | Paid $1,950 |
Yes | Visit |
No payment required. Several of these are complete university courses.
A micro-course that gets you from zero to your first random forest model and Kaggle competition submission in an afternoon. Perfect for people who learn by doing and want immediate, runnable code.
MIT's undergraduate ML course (Fall 2020) with notes, auto-graded exercises, and labs covering perceptrons, gradient descent, neural nets, CNNs, RNNs, and reinforcement learning. Best for learners comfortable with linear algebra who want an MIT-level problem-set experience for free.
Yaser Abu-Mostafa's celebrated course explains why learning works: VC dimension, bias-variance, overfitting, regularization, and validation, alongside core algorithms. Take it if you want the theoretical intuition that most applied courses skip.
Google's refreshed crash course mixes short videos, interactive visualizations, and quizzes across linear and logistic regression, data handling, neural networks, embeddings, LLMs, and production ML. Ideal for engineers who want a practical, fast overview before committing to a longer program.
Microsoft Cloud Advocates' open curriculum walks through regression, classification, clustering, NLP, time series, and RL with Scikit-learn, using real-world datasets from around the globe. A great structured syllabus for self-learners and study groups who prefer notebooks over videos.
The complete recorded lectures of Stanford's graduate ML course, deriving generalized linear models, SVMs, kernels, EM, learning theory, and RL from first principles. Choose this if you want the mathematical depth behind the algorithms, not just how to call them.
Watch everything free; pay only if you want graded assignments and a credential.
Andrew Ng's updated three-course sequence covers regression, classification, neural networks, decision trees, clustering, and recommender systems in Python. Take it as the canonical first ML course if you want a rigorous but gentle on-ramp with hands-on labs.
Structured programs with instructors, cohorts, projects or formal credentials.
The professional-education version of CS229 with graded assignments, course facilitators, and a Stanford record of completion. Choose it when you want the CS229 curriculum with accountability, feedback, and a credential your employer will recognize.
Linear algebra, statistics, calculus, Python, and data analysis for AI work.
Deployment, monitoring, data pipelines, and engineering practices for real AI systems.