Ensure consistency between non-parametric estimates of KM/CIF regarding the addition of each strata for NB, Calibration and Discrimination.

[uriahf]

-[ ] {dcurves} examples

[uriahf]

Seems OK though it's not formal

tim <- c(3.4, 3.4, 0.9, 2.7, 3.1, 5.1, 0.9, 3.4, 4)
delta <- c(1, 0, 1, 0, 1, 1, 1, 1, 0)

surv.obj <- Surv(tim, delta)


cr_cuminc_small <- cmprsk::cuminc(tim, delta)
1 - cmprsk::timepoints(cr_cuminc_small, times = 3)$est


# subsamples
idx <- c(1, 4, 5, 7)

tim_a <- tim[idx]
tim_b <- tim[-idx]

n_a <- length(tim_a)
n_b <- length(tim_b)

delta_a <- delta[idx]
delta_b <- delta[-idx]

cr_cuminc_small_a <- cmprsk::cuminc(tim_a, delta_a)
1 - cmprsk::timepoints(cr_cuminc_small_a, times = 3)$est

cr_cuminc_small_b <- cmprsk::cuminc(tim_b, delta_b)
1 - cmprsk::timepoints(cr_cuminc_small_b, times = 3)$est