relvm: Multiple Estimation Of The Random Effect Latent Variable...

In huangrh/relvm: Random Effect Laten Variable Model For Hospital Compare

Description

Estimate multiple measure groups of the random effect latent variable model

relvm.

Usage

relvm(object, groups = NULL, fit = list(inits = NULL, use_wt = c("step2")),
  file = NULL)

Arguments

object	A mstbl object.
groups	A vector of measure group names. The default is NULL, in which case a vector of all groups will be generated accordingly.
fit	A list of fitting parameters. init: Initial values for mu, fl, and err term in a data frame. fl is the factor loading. They will be initialized generally if it is null. The default is zero for mu and one for the rest. predict: The default is TRUE, which calculate the group score for each hospital. use_wt: A vector of strings, eg. c("step1", "step2"), corresponding to step 1 (optimize the model parameters) and step 2 (predict the group score) in the latent variable model, respectively. For example, fit = list(use_wt = "step2") tells the program to ignore the weight in step 1 and only use the weight in step 2.
file	File name in a rds format to cache the results on hard drive.

Value

An list of S3 object of class "relvms" with estimated parametes.

See Also

mstbl

Examples

# require(rstarating)
# To fit just one group: 'outcome_mort'
fit <- relvm(mstbl(cms_star_rating_input_2017dec),groups="outcome_mort")

# use weight in step 2 only (by default, weigt are used in step 1 and step 2)
fit <- relvm(mstbl(cms_star_rating_input_2017dec),groups="outcome_mort", fit=list(use_wt="step2"))

# Use weight in step 1 only (for test purpose only).
fit <- relvm(mstbl(cms_star_rating_input_2017dec),groups="outcome_mort", fit=list(use_wt="step1"))

huangrh/relvm documentation built on May 17, 2019, 9:10 p.m.