Returns the powers of the fractional polynomial model which has the lowest deviance score.

Return the best powers for a given degree

Usage

find_best_fp_powers(data, p, mc, degree, link = "logit")

Arguments

data

the input data frame, must either have columns for `age`, `pos`, `tot` (for aggregated data) OR `age`, `status` (for linelisting data)

p

a powers sequence to be tested.

mc

indicates if the returned model should be monotonic.

degree

the maximum degree (i.e. number of power terms) to search for the best model. Recommended to be <= 2.

link

the link function. Defaulted to "logit".

Value

list of 3 elements:

p

The best power for fp model.

deviance

Deviance of the best fitted model.

model

The best model fitted