ltsspca (version 0.1.0)

ltspca: Principal Component Analysis Based on Least Trimmed Squaers (LTS-PCA)

Description

the function that computes LTS-PCA

Usage

ltspca(x, q, alpha = 0.5, b.choice = NULL, tol = 1e-06, N1 = 3,
  N2 = 2, N2bis = 10, Npc = 10)

Arguments

x

the input data matrix

q

the dimension of the PC subspace

alpha

the robust parameter which takes value between 0 to 0.5, default is 0.5

b.choice

intial loading matrix; by default is NULL and the deterministic starting values will be computed by the algorithm

tol

convergence criterion

N1

the number controls the updates for a without updating b in the concentration step

N2

the number controls outer loop in the concentration step

N2bis

the number controls the outer loop for the selected b

Npc

the number controls the inner loop

Value

the object of class "ltspca" is returned

b

the unnormalized loading matrix

mu

the center estimate

ws

if the observation in included in the h-subset ws=1; otherwise ws=0

best.cand

the method which computes the best deterministic starting value in the concentration step

References

Cevallos Valdiviezo, H., Van Aelst, S. (2019), `` Fast computation of robust subspace estimators'', Computational Statistics & Data Analysis, 134, 171--185.

Examples

# NOT RUN {
ltspcaM <- ltspca(x = x, q = 2, alpha = 0.5)
# }