rdec: regression with damped exponential correlation

In RDEC: RDEC -- regression with damped exponential correlation

Description

ML fitting of a multivariate gaussian model for longitudinal data that generalizes AR(1), CS, and MA(1) covariance structures

Usage

rdec(formula, id, S, data = sys.parent(), subset, 
   na.action = na.fail, omega.init = c(0, 0), 
   omega.low = c(0.001, 0), omega.high = c(0.95, 0.95), 
   ltol = 0.01, contrasts = NULL, verbose=FALSE)

Arguments

formula	standard linear modeling formula
id	vector of cluster discriminators
S	vector of observation times
data	resolve model formula elements in this data frame
subset	logical subsetting vector
na.action	action to take for NA in formula element
omega.init	starting value for covariance parameter vector (dim=2)
omega.low	lower bound for covariance parameter vector
omega.high	upper bound for covariance parameter vector
ltol	tolerance for convergence of likelihood function
contrasts	probably not used
verbose	print -2 likelihood value per iteration

Details

In the example, it is noted that 'convergence is suspect'. It may be desirable to use other starting points, or to check convergence using a profiling procedure, when such messages are encountered.

Value

a structure with model fitting results

Author(s)

VJ Carey <[e-mail]>

References

Munoz Carey Rosner et al, Biometrics 1992

See Also

gee, geepack, nlme, reshape

Examples

data(rdtest)
summary( rdec(rdec.y~rdec.x, id=rdec.id, S=rdec.s, omega.init = c(.5,.5), omega.low=c(.01,.01),
   omega.high=c(.95, .95), data=rdtest ))

RDEC documentation built on May 2, 2019, 6:28 p.m.