mdprefmx {lazy.mdpref}

R Documentation

Finite Mixture Model for MDPREF

Description

Finite Mixture Model for MDPREF

Usage

mdprefmx(
  f,
  n,
  ij,
  subj,
  ndim = 2,
  nclass = 2,
  alpha = 1,
  init = 2,
  X = NULL,
  B = NULL,
  rho = NULL,
  estX = 1,
  estB = 1,
  estrho = 1,
  minp = 1e-07,
  lmax = 10,
  eps = 1e-06,
  lmax2 = 3,
  print = 1,
  plot = 0
)

Arguments

f	vector consisting of the # of times that the left stimuli was chosen out of n trials.
n	vector consisting of the # of trials.
ij	matrix indicating the stimulus pair.
subj	vector indicating the subject.
ndim	# of dimensions
nclass	# of laten classes
alpha	hyper-parameter vector of size nclass for Dirichlet prior of rho
init	method for initial configuration = 0 to use random number = 1 to use Thurstome Case V, PCA and hclust = 2 to use mdprefml, PCA and hclust
X	initial configuration, if any
B	initial configuration, if any
rho	initial configuration, if any
estX	= 0 to skip the estimation of X and fix it to the value given in X.
estB	= 0 to skip the estimation of B and fix it to the value given in B.
estrho	= 0 to skip the estimation of rho and fix it to the value given in rho.
minp	minimum value of the probability
lmax	max # of iterations
eps	criterion for convergence
lmax2	max # of iterations for the inner iteration for X and Beta
print	= 1 to print the result
plot	= 1 to plot the result

Details

The quartet ( subj, n, f, ij ) contains the result of the paired comparison data.
Thoese four objects has the same length or # of rows.
The k-th elements of the quartet indicates that
subj[k] preferred stimulus ij[k,1] over ij[k,2] f[k] times when exposed to the pair n[k] times.
The paired comparizon for each subject does no have to be complete.

Examples

ij <- cbind( data_comp$ij.1, data_comp$ij.2)
mdprefmx( f=data_comp$f, n=data_comp$n, ij=ij, subj=data_comp$subj, plot=1 )

ij <- cbind( data_miss$ij.1, data_miss$ij.2)
mdprefmx( f=data_miss$f, n=data_miss$n, ij=ij, subj=data_miss$subj, plot=1 )

---