mvgam (version 1.1.594)

series_to_mvgam: Convert timeseries object to format necessary for mvgam models

Description

This function converts univariate or multivariate time series (xts or ts objects) to the format necessary for mvgam.

Usage

series_to_mvgam(series, freq, train_prop = 0.85)

Value

A list object containing outputs needed for mvgam, including 'data_train' and 'data_test'

Arguments

series

xts or ts object to be converted to mvgam format

freq

integer. The seasonal frequency of the series

train_prop

numeric stating the proportion of data to use for training. Should be between 0.25 and 0.95

Examples

# A ts object example
data("sunspots")
series <- cbind(sunspots, sunspots)
colnames(series) <- c('blood', 'bone')
head(series)
series_to_mvgam(series, frequency(series), 0.85)

# An xts object example
library(xts)
dates <- seq(as.Date("2001-05-01"), length = 30, by = "quarter")

data <- cbind(
  c(gas = rpois(30, cumprod(1 + rnorm(30, mean = 0.01, sd = 0.001)))),
  c(oil = rpois(30, cumprod(1 + rnorm(30, mean = 0.01, sd = 0.001))))
)

series <- xts(x = data, order.by = dates)
colnames(series) <- c('gas', 'oil')
head(series)
series_to_mvgam(series, freq = 4, train_prop = 0.85)