Today Tsengel Nergui showed us how she used Principal Component Analysis in her atmospheric research. The script and data provided shows an excellent example of PCA application. Tsengel discusses not only int interpretation of the results, but also some of the standardization that one can do prior to PCA.
In the discussion portion of the session we talked about how a conceptual understanding of PCA can be broken into two philosophies: calculating the eigenvalues or focusing on the dissimilarity matrix. Both lead to the same place but some researchers may find one or the other strategy more compelling. PCA, and indeed other multivariate apraoches in R, are very clearly explained in Manly’s Multivariate Statistical Methods: A Primer. The 4th edition has a website that includes example data and script for R. Another good resource is the R package vegan.
In addition to discussin PCA, we also discussed loading jpegs in R. This is very simple to do with the jpeg package.
This talk will require the following packages:
library(stats) library(plyr) # plyr must be called before dplyr library(dplyr) library(ggplot2) jpeg rasterImage
Necessary script and data below: