hds (version 0.8.1)

finda: Estimate the time-varying coefficients from a local-in-time Cox model

Description

finda estimates the time-varying coefficients beta(t) at a single time from a local-in-time Cox model. Think of it as a Cox model where the the coefficients are allowed to vary with time. Further details can be found in Cai and Sun (2003) and Tian et al. (2005).

Usage

finda(tt, times, status, covars, start = rep(0, ncol(covars)), h = 400, ...)

Arguments

tt

Time to estimate beta(t) at

times

A vector of observed follow up times.

status

A vector of status indicators, usually 0=alive, 1=dead.

covars

A matrix or data frame of numeric covariate values, with a column for each covariate and each observation is on a separate row.

start

A vector of length p of starting values to be passed to optim for the numerical optimization procedure. p is the number of covariates. Defaults to all zeroes.

h

A single value on the time scale representing the bandwidth to use.

...

Additional parameters to pass to optim.

Value

A vector of length p, where p is the number of covariates. The vector is the estimated beta(t) from the local-in-time Cox model at time tt.

Details

The naming of the function finda stands for "find a(t)", where "a(t)" is the notation used in Cai and Sun (2003) to represent the time-varying Cox model coefficients. We also refer to "a(t)" as "beta(t)" through the documentation.

The user typically will not interact with this function, as finda is wrapped by hdslc.

References

Cai Z and Sun Y (2003). Local linear estimation for time-dependent coefficients in Cox's regression models. Scandinavian Journal of Statistics, 30: 93-111. doi:10.1111/1467-9469.00320

Tian L, Zucker D, and Wei LJ (2005). On the Cox model with time-varying regression coefficients. Journal of the American Statistical Association, 100(469):172-83. doi:10.1198/016214504000000845