beyondWhittle (version 1.0)

ar_screeType: Negative log likelihood values for scree-type plots

Description

(Approximate) negative maximum log-likelihood for for different autoregressive orders to produce scree-type plots.

Usage

ar_screeType(data, order.max, method = "yw")

Arguments

data

numeric vector of data

order.max

maximum autoregressive order to consider

method

character string giving the method used to fit the model, to be forwarded to stats::ar

Value

a data frame containing the autoregressive orders p and the corresponding negative log likelihood values nll

Details

By default, the maximum likelihood is approximated by the Yule-Walker method, due to numerical stabililty and computational speed. Further details can be found in the simulation study section in the referenced paper.

References

C. Kirch et al. (2017) Beyond Whittle: Nonparametric Correction of a Parametric Likelihood With a Focus on Bayesian Time Series Analysis <arXiv:1701.04846>

Examples

# NOT RUN {
###
### Interactive visual inspection for the sunspot data
###

data <- sqrt(as.numeric(sunspot.year))
data <- data <- data - mean(data)

screeType <- ar_screeType(data, order.max=15)

# Determine the autoregressive order by an interactive visual inspection of the scree-type plot
plot(x=screeType$p, y=screeType$nll, type="b")
p_ind <- identify(x=screeType$p, y=screeType$nll, n=1, labels=screeType$p)
print(screeType$p[p_ind])
# }