Spectral Clustering using Diffusion Maps with Eigengap Heuristic

Applies K-means clustering on the diffusion coordinates. If the number of clusters is not provided, it automatically estimates the optimal number of clusters (k) using the Eigengap heuristic (finding the largest drop between consecutive eigenvalues).

Usage

cdiff_kmeans(x, centers = "auto", max_k = 10, nstart = 25)

Arguments

x

A 'diffmap' object returned by `run_diffusion()` or `run_diffusion_sparse()`.

centers

Integer or "auto". The number of clusters. If "auto" (default), optimal k is estimated.

max_k

Integer. Maximum number of clusters to consider when `centers = "auto"` (default 10).

nstart

Integer. Number of random sets to choose for K-means (default 25 for stability).

Value

A standard 'kmeans' list object containing the clustering results.