Algorithm Selector

A real branching decision tool over this site's own Machine Learning pages — answer a few questions about your problem (regression vs. classification vs. clustering, linear vs. not, interpretability vs. raw accuracy) and get a specific recommendation with a real reason, not just a name.

What to Try

• Answer "Predict a number," then "No / not sure, or I need maximum accuracy," then "No, I care most about raw accuracy" — you'll land on Gradient Boosting, the practical default for tabular regression when interpretability isn't the priority.
• Try the same first two answers but pick "Yes, interpretability matters" instead — same problem type, different constraint, different (correct) answer: Decision Tree.
• Answer "Find groups in unlabeled data," then "No" (don't know cluster count), then "Yes" (irregular shapes / need outlier detection) — DBSCAN/HDBSCAN, not K-Means, because K-Means structurally can't do either of those things.
• Use Back to change one answer partway through and watch the recommendation change — a concrete way to see which specific constraint is actually driving which specific choice.

Back to Visual Lab Overview.