The 2020 ML Map That Still Mostly Works

What it does This repository is a curated visual roadmap that breaks machine learning into five connected territories: identifying problems, following a process, picking tools, understanding the math, and finding resources. It functions as a curriculum map meant to show how the pieces fit together rather than teaching any single topic in depth. The project offers a static overview image, a separate interactive web version, and a video walkthrough that the author notes runs longer than most feature films.

The interesting bit Most ML resources zoom in on algorithms; this one deliberately zooms out to the full lifecycle. The author titles it a “2020” roadmap and openly claims it is only “still 90% valid for 2023,” a rare admission of decay in a discipline that usually pretends its guides are timeless. That honesty makes the map more useful, not less.

Key highlights

• Five-part structure covering the full lifecycle: Problems, Process, Tools, Mathematics, and Resources
• Interactive version hosted at dbourke.link/mlmap
• Video walkthrough that the README describes as “longer than most movies”
• Credits Daniel Formoso’s earlier machine-learning mindmaps as direct inspiration
• 7,873 stars suggest it has become a popular orientation guide for newcomers

Caveats

• Explicitly dated to 2020; the README itself warns it is only “90% valid for 2023” without specifying which 10% has expired
• It is a visual reference and curriculum guide, not a code repository, so expect no executable examples or libraries

Verdict Grab this if you are a developer drowning in disconnected ML tutorials and need to see how the field fits together. Skip it if you are already shipping models and just want code, not cartography.