Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!
Get Started for Free

crps

Summarizing prediction error curves


Description

Computes the cumulative prediction error curves, aka integrated Brier scores, in ranges of time.

Usage

crps(object, models, what, times, start)

Arguments

object

An object with estimated prediction error curves obtained with the function pec

models

Which models in object$models should be considered.

what

The name of the entry in x to be cumulated. Defauls to PredErr Other choices are AppErr, BootCvErr, Boot632, Boot632plus.

times

Time points at which the integration of the prediction error curve stops.

start

The time point at which the integration of the prediction error curve is started.

Details

The cumulative prediction error (continuous ranked probability score) is defined as the area under the prediction error curve, hence the alias name, ibs, which is short for integrated Brier score.

Value

A matrix with a column for the crps (ibs) at every requested time point and a row for each model

Author(s)

Thomas A. Gerds tag@biostat.ku.dk

References

E. Graf et al. (1999), Assessment and comparison of prognostic classification schemes for survival data. Statistics in Medicine, vol 18, pp= 2529–2545.

Gerds TA, Cai T & Schumacher M (2008) The performance of risk prediction models Biometrical Journal, 50(4), 457–479

See Also

Examples

set.seed(18713)
library(prodlim)
library(survival)
dat=SimSurv(100)
pmodel=coxph(Surv(time,status)~X1+X2,data=dat,x=TRUE,y=TRUE)
perror=pec(list(Cox=pmodel),Hist(time,status)~1,data=dat)

## cumulative prediction error
crps(perror,times=1) # between min time and 1
## same thing:
ibs(perror,times=1) # between min time and 1
crps(perror,times=1,start=0) # between 0 and 1
crps(perror,times=seq(0,1,.2),start=0) # between 0 and seq(0,1,.2)

pec

Prediction Error Curves for Risk Prediction Models in Survival Analysis

v2020.11.17
GPL (>= 2)
Authors
Thomas A. Gerds
Initial release

We don't support your browser anymore

Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.