Principal Component Analysis compute function

[grigory93]

The function computePca would return prcomp-class list (see prcomp) after performing PCA in Aster on Aster table. The flow and implementation is similar to computeKmeans implementation.

[grigory93]

R function prcomp returns a list of class "prcomp" that contains:

• sdev: the standard deviations of the principal components. Components are sorted by these values (which indicate imporance)
• rotation: corresponds to what Aster PCA function computes
• center: logical if centering was used
• scale: logical if scaling was used
• x: the value of the rotated data (we won't support it for now)

Compared to Aster implementation the following are missing parts that function computePca should offer:

• return the list compatible with prcomp being both classes "prcomp" and "toapca"
• just like prcomp (arguments center and scale) support centering and scaling (in any combination) using Aster scale functions (ScaleMap and Scale)
• sdev values. I believe it's possible to calculate somehow - needs checking.

Please, approach this requirements rather critically and correct as appropriate.

[grigory93]

Existing workflow implemented in computeKmeans can serve as a template for computePca