get_prior_weights: Compute priors on the mixing weights of each candidate latent...

In hillarykoch/CLIMB: Composite LIkelihood eMpirical Bayes

Compute priors on the mixing weights of each candidate latent class.

Description

This is the third step of the CLIMB procedure. It is useful because one can eliminate latent classes whose prior weights are too small. This function is parallelizable.

Usage

get_prior_weights(reduced_classes, fits, parallel = FALSE, ncores = 20, delta = NULL)

Arguments

reduced_classes	Matrix output from get_reduced_classes.
fits	Fits output from get_pairwise_fits.
parallel	Boolean. Should this function be parallelized?
ncores	If parallel=TRUE, how many cores should this be parallelized over?
delta	Which value of integer tuning parameters delta should be checked? Defaults to 0:choose(D,2)

Value

List of vectors. Each element of the list is a vector corresponds to the computed prior weights for a given value delta.

Author(s)

[e-mail]

Examples

data(sim)
data(fits)

#------------------------------------------------------------
# Not run:
#
# D <- ncol(sim$data)  
# reduced_classes <- get_reduced_classes(fits, D, "lgf.txt")
# prior_weights <- get_prior_weights(reduced_classes, fits, parallel = FALSE)
# plot(
#    seq_along(prior_weights)-1,
#    sapply(prior_weights, function(X)
#    sum(X * n > D)),
#    ylab = "number of retained classes",
#    xlab =  expression(delta))
#------------------------------------------------------------

hillarykoch/CLIMB documentation built on Oct. 24, 2022, 4:27 a.m.