Patterns in Multivariate Data

Thus far, we have restricted our analysis to one-dimensional time series data sets. However, many climate and geophysical data sets are multi-dimensional and exhibit variability in both time and in space. In the final section of this courseware, we will explore one of the most common methods for describing the variability of these data sets, principal component analysis (PCA).

We will start with an example of the El Niño-Southern Oscillation to help us develop an intuitive sense of PCA is all about. Next, we will review some of the matrix algebra and eigenanalysis concepts needed to understand how PCA works. Finally, we will learn how to conduct PCA in practice using python.