gendataWmm {lazy.procrustes} | R Documentation |
Generation of Procscal Data
gendataWmm(ns, nv, nr, genm = 0, genW = 0, genT = 1, genW2 = 0, genmu = 0, mu0 = 0, genD1 = 0, genD2 = 0, eqm = 0, eqW = 0, eqT = 0, eqW2 = 0, diagW2 = 1, eqD1 = 0, eqD2 = 0, genV = 0, diagV = 1, pcA = 0, errstd = 0, pmiss = 0, minWD = 0.2)
ns |
# of matrices to be generated. |
nv |
# of rows of the matrices to be rotated. |
nr |
# of columns of the matrices to be rotated. |
genm |
= 1 to generate m vectors, the column mean |
genW |
= 1 to generate scalar W matrices, the left weight |
genT |
= 0 to skip the generation of T matrices, the rotation |
genW2 |
= 1 to generate W2 matrices, the right weigt |
genmu |
= 1 to generate mu matrix |
mu0 |
The common value of mu |
genD1 |
= 1 to generate D1 matrices, |
genD2 |
= 1 to generate D2 matrices, |
eqm |
= 1 to generate common m |
eqW |
= 1 to generate common W |
eqT |
= 1 to generate common T |
eqW2 |
= 1 to generate common W2 |
diagW2 |
= 0 to make W2 non-diagonal |
eqD1 |
= 1 to generate common WD1 |
eqD2 |
= 1 to generate common D2 |
genV |
= 1 to generate V matrix |
diagV |
= 0 to make V non-diagonal |
pcA |
= 1 to make the column variances of A equal |
errstd |
std of errors to be added to final C matrix |
pmiss |
= proportion of missing elements of C matrix |
minWD |
= minimum value of W and D matrices |
This program generates A, mk, Tk, Wk, W2k, D1k, D2k matrices
so that the k-th target matrix Ck is created as
Ck = inv(D1k) ( A Wk Tk W2k + 1 mk' ) inv(D2k) + Ek, k=1,2,...,ns,
where Disp(Ek) = errstd^2*solve(V).
When genmu=1, we set genm=0; genD1=0; genD2=0; genW2=0; genV=0 and
muk = mu0 + A uk
will be generated via uk where muk is nv x 1 and uk is nr x 1.
For examples, see procotWKS or procscal.
A list of all the matrices and the parameter values.