Pca

Principal-component decomposition of a symmetric covariance or correlation matrix by the cyclic Jacobi eigenvalue algorithm — the standard first step in a term-structure factor analysis (level / slope / curvature) and in reducing a large correlation matrix to a handful of driving factors.

eigenvalues are in descending order; column i of loadings is the unit eigenvector for eigenvalues[i]; explainedVariance[i] is that eigenvalue as a fraction of the trace.