get_survroc: Calculate survival ROC

In csetraynor/rstanhaz: Bayesian Survival Models (rstanhaz)

Description

These functions calculate the survival analysis metric measured of a system compared to a hold-out test set. The measurement and the "truth" have a survival time and a censoring indicator 0/1 indicating if the event result or the event.

Usage

get_survroc(holdout, fit, time = "time", status = "status",
  ncores = ncores)

Arguments

data	For the default functions, a datframe containing survival (time), and status (0:censored/1:event), and the explanatory variables.
mod	Coxph model object fitted with coxph (survival).

Details

The Brier score is defined as the squared distance between the expected survival probability and the observed survival. Therefore, it measures the discrepancy between observation and model-based prediction.

The integrated Brier Score summarises the Brier Score over the range of observed events.Similar to the original Brier score [40] the iBrier: ranges from 0 to 1; the model with an out-of-training sample value closer to 0 outperforms the rest.

Value

A tdBrier object

Author(s)

Carlos S Traynor

References

Ulla B. Mogensen, Hemant Ishwaran, Thomas A. Gerds (2012). Evaluating Random Forests for Survival Analysis Using Prediction Error Curves. Journal of Statistical Software, 50(11), 1-23. URL http://www.jstatsoft.org/v50/i11/.

See Also

[iBrier]

Examples

require(survival)
require(dplyr)
data(lung)
lung <- lung %>%
mutate(status = (status == 2))

mod <- coxph(Surv(time, status)~ age, data = lung)

tdbrier <- get_tdbrier(lung, mod)
integrate_tdbrier(tdroc)

csetraynor/rstanhaz documentation built on May 9, 2019, 8:14 a.m.