Suggestion: ggcoxfunctional works with coxph object not formulas

[pbiecek]

Would you consider a change in the design of ggcoxfunctional() function?

Right now it takes the formula and data, and is calling coxph function internally.
But it may be more flexible to separate the modelling and variable diagnostic.

The suggestion is to send an argument 'fit' instead of formula and data objects,
you can extract both formula and data from the 'fitobject'.

It will be also consistent with ggcoxzph, ggcoxdiagnostics and ggsurvplot functions.

[MarcinKosinski]

Thanks for the suggestion.
I did not realize this inconsistency.

[kassambara]

The design of the ggcoxfunctional()is now consistent with the other functions.

library(survival)
fit <- coxph( Surv(futime, death) ~ mspike + log(mspike) + I(mspike^2) +
                     age + I(log(age)^2) + I(sqrt(age)), data = mgus)

library(survminer)
ggcoxfunctional(fit)

[pbiecek]

Great!