iCARH.plotBeta {iCARH}

Postprocess and plot model parameters

Description

Group of functions to postprocess and plot model parameters of interest, compute WAIC (Watanabe-Akaike Information Criterion) and MADs (Mean Absolute Deviation) for posterior predictive checks and check normality assumptions.

Usage

iCARH.plotBeta(fit, indx = TRUE, indy = TRUE)

iCARH.plotARCoeff(fit, indx = TRUE)

iCARH.plotTreatmentEffect(fit, indx = TRUE)

iCARH.plotPathwayPerturbation(fit, path.names, indpath = TRUE)

iCARH.plotDataImputation(fit, indx = T, indy = T, plotx = T, ploty = T, ...)

iCARH.checkRhats(fit)

iCARH.checkNormality(fit)

iCARH.waic(fit)

iCARH.mad(fit)

Arguments

fit

object returned by iCARH.model

indx

vector to specify X variables to plot. Selects all variables of X by default.

indy

vector to specify Y variables to plot. Selects all variables of Y by default.

path.names

pathway names

indpath

vector to specify pathways to plot. Selects all pathways by default.

plotx

plot X data imputation?

ploty

plot Y data imputation?

...

passed to ggplot2::geom_violin

Value

the iCARH.plot[*] functions return a ggplot graph object. iCARH.checkNormality returns the normalized data. iCARH.waic and iCARH.mad return corresponding waic (scalar) and mad (vector of J*(J+1)/2) values. iCARH.checkRhats checks model convergence.

Functions

• iCARH.plotBeta: Plot boxplots of posterior densities of \beta coefficients.

• iCARH.plotARCoeff: Plot boxplots of posterior densities of theta (time effect) coefficients.

• iCARH.plotTreatmentEffect: Plot boxplots of posterior densities of treatment effect coefficients.

• iCARH.plotPathwayPerturbation: Plot posterior densities of pathway perturbation parameters

• iCARH.plotDataImputation: Plot imputed values

• iCARH.checkRhats: check model convergence and return Rhat coefficients

• iCARH.checkNormality: Check normality assumptions. Returns normalized data and performs quantile-quantile plot

• iCARH.waic: Compute Watanabe-Akaike Information Criterion (WAIC)

• iCARH.mad: Compute MADs (Mean Absolute Deviation) between true covariance matrix and inferred covariance matrix for posterior predictive checks

Examples

data.sim = iCARH.simulate(4, 10, 14, 8, 2, path.probs=0.3, Zgroupeff=c(0,4),
beta.val=c(1,-1,0.5, -0.5))
XX = data.sim$XX
Y = data.sim$Y
Z = data.sim$Z
pathways = data.sim$pathways

rstan_options(auto_write = TRUE)
options(mc.cores = 2)
fit = iCARH.model(XX, Y, Z, groups=rep(c(0,1), each=5), pathways, 
control = list(adapt_delta = 0.99, max_treedepth=10), iter = 2, chains = 2)
if(!is.null(fit$icarh))
gplot = iCARH.plotBeta(fit, indx=1:3, indy=1:2)