FactorCopulaModel (version 0.1.1)

mvtBifact: MLE for multivariate normal/t with a bi-factor or nested factor correlation structure

Description

MLE for the bi-factor or nested factor structure for multivariate normal/t

Usage

mvtBifact(tdata,start,grsize,df,prlevel=2,model="bifactor",mxiter=100)

Value

nlm object with ($code,$estimate,$gradient,$iterations,$minimum)

Arguments

tdata

nxd matrix of t-scores or z-scores

start

vector of length 2*d with starting values of partial correlations values for correlations of observed Z_[gj] and common latent V_0 go first, then partial correlations of Z_[gj] and V_g given V_0 (j in group g)

grsize

vector of group sizes for bi-factor model

df

degrees of freedom parameter >0

prlevel

print.level for nlm()

model

"bifactor" or "nestfactor" nested-factor is reduced model with fewer parameters

mxiter

maximum number of iterations for nlm()

Details

Note the minimum nllk can be the same for different parameter vectors if some group size values are 1 or 2.

Examples

# \donttest{
data(rainstorm)
udat = rainstorm$uprecip
d = ncol(udat)
grsize = rainstorm$grsize
df = 10
tdata = qt(udat,df)
bif = mvtBifact(tdata, c(rep(0.8,d),rep(0.2,d)), grsize, df=df,
prlevel=1, model="bifactor", mxiter=100)
#
# nested-factor: parameters for group latent linked to global latent
# come in the first tree of the 2-truncated vine.
nestf = mvtBifact(tdata, c(0.7,0.7,0.7,rep(0.85,d)), grsize, df=df,
prlevel=1, model="nestfactor", mxiter=100)
# doesn't converge properly, group2 latent matches global latent
#
st1 = rep(0.7,d)
out1t = mvtPfact(tdata,st1,pfact=1,df=df,prlevel=1)
st2 = rep(0.7,2*d)
out2t = mvtPfact(tdata,st2,pfact=2,df=df,prlevel=1)
st3 = c(rep(0.7,grsize[1]),rep(0.1,d),rep(0.7,grsize[2]),rep(0.1,d),rep(0.7,grsize[3]))
out3t = mvtPfact(tdata,st3,pfact=3,df=df,prlevel=1)
cat(bif$minimum, nestf$minimum, out1t$minimum, out2t$minimum, out3t$minimum,"\n")
# }