Build a linear parameter constraint

tda_rate takes constraints in TDA's notation, where parameters are numbered in the order they are estimated: "b3 - b10 = 0" makes the third and tenth equal. This builds such a string from a fitted model's coefficient names, so the numbering does not have to be counted out by hand.

Usage

tda_constrain(fit, ...)

Arguments

fit

a fit from tda_rate, used for its coefficient names.

...

constraints, each a character vector of two coefficient names to be held equal, or a formula like EDU ~ PRES.

Value

A character vector of constraints for tda_rate.

See also

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

Examples

d <- tda_rrdat(states = 4)
f <- tda_rate(Surv(TFP, DES) ~ EDU + PRES, d)
tda_constrain(f, c("0->1: EDU", "0->2: EDU"))
#> [1] "b2 - b5 = 0"