loglikLOOCVVAR2: Leave-one-out (minus) cross-validated log-likelihood of...

Description Usage Arguments Value Note Author(s) References See Also Examples

View source: R/loglikLOOCVVAR2.r

Description

Evaluation of the (minus) leave-one-out cross-validated log-likelihood of the VAR(2) model for given choices of the ridge penalty parameters (λ_{a1}, λ_{a2} and λ_{ω} for the lag one autoregression coefficient matrix \mathbf{A}_1, lag two autoregression coefficient matrix \mathbf{A}_2 of time-varying covariates, and the inverse error covariance matrix \mathbf{Ω}_{\varepsilon} (=\mathbf{Σ_{\varepsilon}^{-1}}), respectively). The functions also works with a (possibly) unbalanced experimental set-up. The VAR(2)-process is assumed to have mean zero.

Usage

1
loglikLOOCVVAR2(lambdas, Y, unbalanced=matrix(nrow=0, ncol=2), ...)

Arguments

lambdas

A numeric of length three, comprising positive values only. It contains the ridge penalty parameters to be used in the estimation of \mathbf{A}_1, \mathbf{A}_2 and the precision matrix of the errors.

Y

Three-dimensional array containing the response data. The first, second and third dimensions correspond to variates, time and samples, respectively. The data are assumed to be centered covariate-wise.

unbalanced

A matrix with two columns, indicating the unbalances in the design. Each row represents a missing design point in the (time x individual)-layout. The first and second column indicate the time and individual (respectively) specifics of the missing design point.

...

Other arguments to be passed to the ridgeVAR2-function.

Value

A numeric of length one: the minus (!) LOOCV log-likelihood.

Note

The minus LOOCV log-likelihood is returned as standard optimization procedures in R like nlminb and constrOptim minimize (rather then maximize). Hence, by providing the minus LOOCV log-likelihood the function loglikLOOCVVAR2 can directly used by these optimization procedures.

Author(s)

Wessel N. van Wieringen <[email protected]>

References

Miok, V., Wilting, S.M., Van Wieringen, W.N. (2017), “Ridge estimation of network models from time-course omics data”, submitted.

See Also

ridgeP and ridgeVAR2.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
# set dimensions (p=covariates, n=individuals, T=time points)
p <- 3; n <- 12; T <- 10

# set model parameters
SigmaE <- diag(p)/4
A1 <- createA(p, topology="clique", nonzeroA=0.1, nClique=1)
A2 <- createA(p, topology="hub", nonzeroA=0.1, nHubs=1)

# generate data
Y <- dataVAR2(n, T, A1, A2, SigmaE)

## determine optimal values of the penalty parameters
## Not run: optLambdas <- constrOptim(c(1,1, 1), loglikLOOCVVAR2, gr=NULL, 
## Not run:               ui=diag(3), ci=c(0,0,0), Y=Y,
## Not run:               control=list(reltol=0.01))$par 

# ridge ML estimation of the VAR(2) parameter estimates with 
# optimal penalty parameters
optLambdas <- c(0.1, 0.1, 0.1)
ridgeVAR2(Y, optLambdas[1], optLambdas[2], optLambdas[3])

ragt2ridges documentation built on Nov. 21, 2017, 5:06 p.m.