compare.surv.mods: Comparing Piecewise Exponential model with other Parametric...

In Philip-Cooney/PiecewiseChangepoint: Piecewise Exponential Models with Unknown Change-points

Comparing Piecewise Exponential model with other Parametric models

Description

Compares the piecewise exponential model with 6 other standard parameteric models (see fit_surv_models()). Note that exponential is a special case of the piecewise exponential, however, it is refit in JAGS to highlight the difference in statistical fit. This functions computes the individual log-likelihood for the piecewise exponential model (see get.loglik()) and compares the Widely Applicable Information Criterion (using the loo::loo-package) and Pseudo-Marginal Likelihood (PML) with the other standard parametric models.

Usage

compare.surv.mods(
  object,
  max_predict = 10,
  chng.num = "all",
  plot.best = 3,
  n.iter.jags = 2000,
  n.thin.jags = NULL,
  n.burnin.jags = NULL,
  gof = "WAIC",
  inc_waic = TRUE,
  km_risk = 0.1,
  gmp_haz_df = NULL,
  gpm_post_data = TRUE,
  col_km = "black",
  final_chng_only = FALSE
)

Arguments

object	of class "changepoint".
max_predict	maximum survival time to be predicted for the survival models. Default is 10, however, depending on the timescale this should be changed.

Value

A list of with the following items:

• model.comp: A dataframe with the PML and WAIC for the piecewise exponential model and the six parametric models fitted by JAGS.

• jags.models: A list containing the posterior simulations of the 6 JAGS models (fit using the R2jags:jags function).

• plot_Surv_all: A ggplot with the posterior mean survival probabilities for the time specified by max_predict.

Philip-Cooney/PiecewiseChangepoint documentation built on Sept. 10, 2023, 9:49 p.m.