CURVA Framework — Clustering fiable

Field notes

CURVA breaks clustering analysis into an explicit evidence chain: initial audit, imputation, outlier treatment, feature handling, representation choice, model comparison, and stability checks. It offers four built-in representations and explores cluster counts from 2 to 6 across four model families. The included examples span a high-dimensional agricultural file and a 30-variable biomedical dataset whose diagnostic labels are withheld, allowing the interface to demonstrate unsupervised analysis rather than supervised classification.