summary.basta: Summarizing and plotting Bayesian Survival Trajectory Analysis (BaSTA) model outputs.

Description

These functions are all generic methods for class basta.

Usage

## S3 method for class 'basta':
summary(object, \dots)
## S3 method for class 'basta':
print(x, \dots)
## S3 method for class 'basta':
plot(x, plot.trace = TRUE, trace.name = "theta", \dots)

Arguments

object

An object of class basta.

plot.trace

A logical argument. If TRUE the raw parameter traces are plotted, else, the predictive intervals for the resulting survival probability and mortality rates are plotted.

trace.name

Character string indicating the set of parameters or posteriors to be plotted. The options are: "theta" to plot the survival model parameters; "gamma" to plot the proportional hazards parameters (if it applies, else plot.ba

...

Additional arguments passed to functions print, summary and plot (see details).

Value

Function summary() outputs the folowing values:
coefficientsA matrix with estimated coefficients (i.e. mean values per parameter on the thinned sequences after burnin), which includes standard errors, upper and lower 95% credible intervals, update rates per parameter (commonly the same for all survival and proportional hazards parameters), serial autocorrelation on the thinned sequences and the potential scale reduction factor for convergence (see Convergence value below).
DICBasic deviance information criterion (DIC) calculations to be used for model selection (Spiegelhalter et al. 2002).
KullbackLeiblerList with Kullback-Leibler discrepancy matrices between pair of parameters for categorical covariates (McCulloch 1989, Burnham and Anderson 2001) and McCulloch's (1989) calibration measure. If only one simulation was ran or if no convergence was reached, then the returned value is Not calculated.
convergenceA matrix with convergence coefficients based on potential scale reduction as described by Gelman et al. (2004). If only one simulation was ran, then the returned value is Not calculated.
modelSpecsModel specifications inidicating the model, the shape and the covariate structure that were specified by the user.
settingsA vector indicating the number of iterations for each MCMC, the burn in sequence, the thinning interval, and the number of simulations that were run.

Details

For objects of class basta, the print function returns three summary elements describing the model and its results, namely: call, run, coefficients and, if convergence was reached, the DIC values for model fit. call describes the basic model used (i.e. exponential, Gompertz, Weibull or logistic), the shape chosen, simple, Makeham or bathtub, the covariate structure chosen, fused, prop.haz or all.in.mort and which covariates (if any) were categorical and which continuous. Argument digits can be used for number formatting (see summary() or signif() for details).

The summary element coefficients prints out the estimated coefficients for all parameters in the model, as well as their standard errors and 95% upper and lower credible intervals. It also includes a measure of serial autocorrelation for each parameter calculated from the thinned parameter chains, an update rate per parameter, and the potential scale reduction factor for each parameter as a measure of convergence (Gelman et al. 2004).

Function summary includes all the previous elements, as well as a summary description of the priors and jump standard deviations for all survival parameters, a calibration of the Kullback-Leibler discrepancy as a measure of parameter similarities for those parameters associated to categorical covariates (McCulloch 1989), and a measure of model fit based on the deviance information criterion (DIC) (Spiegelhalter et al. 2002).

Function plot takes objects of class basta to create trace plots or, if the argument for plot.trace is set to FALSE, it plots estimated survival probabilities and mortality rates with their 95% predictive intervals. If argument plot.trace is set to FALSE, argument xlim can be used to define a range of ages to visualize survival and mortality trends. Also, if logical argument noCI is set to TRUE, credible intervals around survival and mortality are not plotted, leaving only the mean trends. This can be handy when several categorical covariates have been evaluated and the plots get too crowded.

References

Gelman, A., Carlin, J.B., Stern, H.S. and Rubin, D.B. (2004) Bayesian data analysis. 2nd edn. Chapman & Hall/CRC, Boca Raton, Florida, USA.

McCulloch, R.E. (1989) Local model influence. Journal of the American Statistical Association, 84, 473-478.

Spiegelhalter, D.J., Best, N.G., Carlin, B.P. and Van Der Linde, A. (2002) Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society: Series B 64, 583-639.

Examples

Run this code

## Load BaSTA output:
data("sim1Out", package = "BaSTA")

## Print summary output:
summary(sim1Out)

## Plot traces for mortality parameters (theta):
plot(sim1Out)

## Plot traces for proportional hazards parameters (gamma):
plot(sim1Out, trace.name = "gamma")

## Plot traces for recapture probability(ies) (pi):
plot(sim1Out, trace.name = "pi")

## Plot predicted mortality and survival:
plot(sim1Out, plot.trace = FALSE)

## Plot predicted mortality and survival between 2 and 8 years of age:
plot(sim1Out, plot.trace = FALSE, xlim = c(2, 8))

Run the code above in your browser using DataLab