Principal Coordinate Analysis

method
Principal Coordinate Analysis (PCOA) is a distance-based ordination that aims to compress similarity patterns into an ordination diagram.
Published

January 26, 2024

Modified

January 26, 2024

Description

Principal Coordinate Analysis (PCOA), also known as classical (metric) multidimensional scaling, is a statistical technique used to visualise and analyse (dis)similarities between a set of objects or observations. PCOA transforms a distance matrix derived from pairwise dissimilarity measures into a set of orthogonal axes (principal coordinates) in a lower dimensional space. These axes represent the main directions in the data that account for the maximum variance.

Example

lalashan <- readRDS("lalashan.rds")
library(vegan)

# First calculate distance matrix
dist_m <- vegdist(lalashan$cross, method = "bray")

# Compute the ordination
pcoa_ord <- wcmdscale(dist_m, eig = TRUE)
plot(pcoa_ord)

Alternative functions

  • vegan::wcmdscale()
  • vegan::capscale() if using only one matrix as input
  • stats::cmdscale()
  • ape::pcoa()

Further References