premium_wrapper: Wrapper function to fit profile regression model from package...

In lvhoskovec/mmpack: Methods for multipollutant mixtures analysis

Description

Wrapper function to fit profile regression model from package PReMiuM

Usage

premium_wrapper(
  niter,
  nburn,
  Y,
  X,
  W,
  scaleY = FALSE,
  varSelectType = "None",
  simnum = NULL,
  priors,
  seed = NULL
)

Arguments

niter	number of iterations through sampler
nburn	number of burn-in iterations
Y	response data
X	exposure data
W	covariate data
scaleY	logical; if TRUE response is centered and scaled before model fit
varSelectType	type of variable selection to be used: "None" or "BinaryCluster", default is "None"
simnum	number of output file (.txt files will be saved in the file path paste0("Premium_output/output",simnum)
priors	prior hyperparameters
seed	random number seed

Value

list with components

• fit: model fit, an object of type "runInfoObj". See profRegr documentation in R package PReMiuM

• risk: mean risk for each subject

• rho: posterior distribution of probability of inclusion for each exposure

• risk.summary: mean, SD, and 0.95 CI of cluster-intercept (risk) for each subject

• exposure.response: cluster means, SD, 0.95 CI, and size, plus mean/SD of repsonse for subjects in each cluster

• fitted vals: fitted values for each subject

• clusters: vector of most optimal clustering

• cluster.summary: array of empirical mean and SD of exposures for individuals assigned to each cluster

• risk distn: distribution of model-averaged cluster-intercepts for each cluster in the best clustering

• groupList: list of which subjects are in which group in the most optimal clustering

• riskProfileObj: object of type "riskProfileObj". See PReMiuM documentation for function "calcAvgRiskAndProfile"

lvhoskovec/mmpack documentation built on Aug. 21, 2021, 4:05 p.m.