Order cluster labels k1, k2, ... k30 rather than k1, k10, k11, ... k2

Cluster ids are "k" plus an integer, and sorting them as text puts k10 before k2. In a 30-cluster legend that is not a cosmetic problem: the reader looks for k7 between k6 and k8 and finds it near the bottom, and colour-to-cluster matching stops being possible at a glance.

Usage

natural_cluster_factor(x)

Arguments

x

Character vector of cluster ids.

Value

x as a factor in natural order.