H(t) = sum d_i/n_i. Better behaved than -log(KM) in small
samples, and the natural scale on which to judge proportional
hazards.
Examples
if (requireNamespace("survival", quietly = TRUE)) {
V <- c(1, 2, 3, 4, 5, 6, 7, 8)
morie_nelson_aalen(V, V)
}
#> $time
#> [1] 1
#>
#> $cumhaz
#> [1] 0.125
#>
#> $se
#> [1] 0.125
#>
#> $surv
#> [1] 0.8824969
#>
