R Graphical Manual

Browse All

Last data update: 2014.03.03

R: Fit Multiplicative AR(1) with Seasonal Process to Data

fit.mar1s

R Documentation

Fit Multiplicative AR(1) with Seasonal Process to Data

Description

Fits Multiplicative AR(1) with Seasonal process model to time series.

Usage

fit.mar1s(x, xreg = NULL, seasonal.fun = seasonal.smooth, ...)

Arguments

`x`	A univariate time series.
`xreg`	A univariate or multivariate time series of external regressors, or `NULL`.
`seasonal.fun`	A function which takes a univariate time series as its first argument and returns the estimated seasonal component.
`...`	Additional arguments passed to `seasonal.fun`.

Value

An object of class "mar1s" with the following components:

`logseasonal`	Estimated log-seasonal figure (a univariate or multivariate time series object).
`logstoch.ar1`	AR(1) with external regressors model fitted for the log-stochastic component (an object of class `"Arima"`).
`logresid.sd`	Standard deviation of the residuals.
`decomposed`	An object of class `"mar1s.ts"` containing decomposed time series (see `compose.mar1s`).

Examples

data(forest.fire, package = "mar1s")
data(nesterov.index, package = "mar1s")

## Simple
mar1s <- fit.mar1s(forest.fire)
plot(mar1s$logseasonal)
confint(mar1s$logstoch.ar1)
mar1s$logresid.sd
resid <- nan2na(mar1s$decomposed$logresid)
qqnorm(resid)
qqline(resid)

## External regressors
mar1s <- fit.mar1s(forest.fire, nesterov.index[, "mean"])
plot(cbind(mar1s$logseasonal, mar1s$logseasonal.r))
confint(mar1s$logstoch.ar1)
resid <- nan2na(mar1s$decomposed$logresid)
qqnorm(resid)
qqline(resid)