Clustering wallets by behavior (transaction frequency, value, token holdings)

This lesson is for enrolled learners

Concept "Clustering" sounds like it requires machine learning, but for a reproducible, citable research report, rule-based clustering in plain SQL is usually the better tool. The approach: compute a handful of per-wallet features — transaction count, average transaction size, number of distinct tokens held, days since last activity — then use CASE WHEN bucket logic or percentile functions (NTILE, PERCENT_RANK)……

Enroll to continue Already enrolled? Sign in