Calculates a Configuration based on the principal components from the selected TableOfReal.

Setting

Number of dimensions

determines the number of dimensions of the resulting Configuration.

Algorithm

We form principal components without explicitly calculating the covariance matrix C = M′ M, where M is the matrix part of the TableOfReal.

1. Calculate the singular value decomposition of M. This results in M = U D V′.

2. The principalComponent_ij = ∑_{k=1..numberOfColumns} M_ik · V_jk.

Remark

The resulting configuration is unique up to reflections along the new principal directions.

Links to this page

• Principal component analysis