stanfitExtended: `stanfitExtended`, an S4 class inherited from the S4 class `stanfit`

Description

Inherits from the class stanfit which is an S4 class defined in the package rstan :

Arguments

Slots

plotdataMRMC: Plot data for MRMC case.

plotdata

This is a data frame with four components which is used to draw curves such as FROC curves and AFROC curves. So, this slot includes the component:

fit@plotdata$x.AFROC,

fit@plotdata$y.AFROC,

fit@plotdata$x.FROC,

fit@plotdata$y.AFROC

where fit is an object of class stanfitExtended.

For example, we can use this slot such as plot(fit@plotdata$x.AFROC, fit@plotdata$y.AFROC ) , where fit is a fitted model object of class stanfitExtended. The author think this slot is not so good since it increase the object size.

dataList

An FROC dataset. Using the dataset, the fitting has done.

studyDesign

A character, e.g., "srsc.per.image", "srsc.per.lesion", according to False Positive Fraction (FPF) is per image or per lesion.

metadata

An additional data calculated from dataList, such as cumulative hits and false alarms,...,etc.

WAIC

A WAIC calculated by the function waic .

convergence

A logical R object TRUE or FALSE. If TRUE, then it means your model is good in the R hat criterion.

PreciseLogLikelihood

This is TRUE or FALSE. If TRUE, then target formulation is used in the stan file. However, non-target formulation has warning for non-linear Jacobian issue. So, the author use target formulations for all .stan files, and thus this slot is now, redandunt.

chisquare

This is a chi square calculated with Expected A Posterior estimates, i.e., the posterior mean estimates. Chi square statistic is $\chi^2 (Data|\theta)$, there are three simple ways to get it.

(1) $ \int \chi^2(Data|\theta ) f(Data|\theta)\pi(\theta|Data)d\theta $

(2) $ \chi^2(Data|\int \theta \pi(\theta|Data)d\theta) $

(3) $ \int \chi^2(Data|\theta ) f(Data|\theta)\pi(\theta|Data)d\theta $

where, $f( Data|\theta )$ denotes a likelihood and $\pi(\theta| Data )$ is a posterior. This slot retains the (2)

Note that this is not calculated by integrating the posterior predictive measure. Do not confuse with the p value calculated with the posterior predicitive measure implemented in the function ppp()

index

This is for programming phase.

Divergences

This is a number of the divergence transitions in the MCMC simulation.

MCMC.Iterations

A MCMC iterations which does not count the burn-in period.

Divergence.rate

A divergence rate, that is the number of the divergence iterations over total MCMC iterations. Burn-in period is not included.

model_name

A slot from the stanfit which is an S4 class defined in the rstan package.

model_pars