The survivor function as a single table

TDA writes one block per group and its examples select a row range out of the flat file with tsel = Case[16,,30]. This returns the whole thing as one long data frame instead, with a group column, so it can be plotted, subset or joined without knowing how many blocks there were.

Usage

tda_survivor(x, conf.int = 0.95)

Arguments

x

a fit from tda_ltb, tda_ple or tda_km.

conf.int

width of the confidence band to add as lower and upper, or NULL for none. TDA has no band of its own – it reports the standard error and leaves the band to the caller – so these are the usual normal limits, clamped to [0, 1].

Value

A data frame with group, time, survivor, std.err, and where available density, rate, lower and upper.

Details

The same call works for tda_ltb and tda_ple: both name their time column time here, whatever the underlying command called it, and both carry survivor and std.err. Where the estimator provides them, density and rate come too.

See also

Other rate models: tda_constrain(), tda_control(), tda_dple(), tda_rates(), tda_split(), tda_transitions(), vcov.tda_fit()

Examples

d <- data.frame(t = c(4, 3, 1, 5, 8, 2), s = c(1, 1, 0, 1, 1, 1),
                g = c(1, 1, 1, 2, 2, 2))
f <- tda_ltb(Surv(t, s) ~ as.factor(g), d, tp = seq(0, 10, 2))
tda_survivor(f)
#>   group time  survivor   std.err   density      rate    lower    upper
#> 1     1    0 1.0000000 0.0000000 0.0000000 0.0000000 1.000000 1.000000
#> 2     1    2 1.0000000 0.0000000 0.2500000 0.3333333 1.000000 1.000000
#> 3     1    4 0.5000000 0.3535534 0.2500000 1.0000000 0.000000 1.000000
#> 4     2    0 1.0000000 0.0000000 0.0000000 0.0000000 1.000000 1.000000
#> 5     2    2 1.0000000 0.0000000 0.1666667 0.2000000 1.000000 1.000000
#> 6     2    4 0.6666667 0.2721655 0.1666667 0.3333333 0.133232 1.000000
#> 7     2    6 0.3333333 0.2721655 0.0000000 0.0000000 0.000000 0.866768
#> 8     2    8 0.3333333 0.2721655 0.1666667 1.0000000 0.000000 0.866768