Rust implementation of prcomp

Runs the singular value decomposition over the matrix x. Assumes that samples = rows, and columns = features.

Usage

rs_prcomp(x, scale, top_pcs)

Arguments

x

Numeric matrix. Rows = samples, columns = features.

scale

Boolean. Shall the columns be variance normalised. (Mean centring will automatically occur.)

top_pcs

Optional integer. Only return the top PCs (under the hood all of them will be calculated). NULL returns all.

Value

A list with:

• scores - The product of x (centred and potentially scaled) with v.

• v - v matrix of the SVD.

• s - Standard deviations of the PCs, i.e. singular values divided by sqrt(nrow(x) - 1).

• scaled - Boolean. Was the matrix scaled.