deseason: Function to deseasonalize a time series.

In gu-stat/gears: GEARS Method for Univariate and Multivariate Time Series Forecasting

Description

This function deseasonalizes a time series using the "multiplicative" decomposition only.

Usage

deseason(ts.data, ts.frequency, forecast.horizon, alpha.level = 0.05)

Arguments

ts.data	A ts object (i.e., a vector of time series observations).
ts.frequency	The frequency of a ts object.
forecast.horizon	Number of periods for forecasting.
alpha.level	The alpha level to be used in the test to detect seasonality. Default is 0.05.

Value

A list with the deseasonalized data ("deseasonTS") and the seasonal component ("seasonalComp").

Examples

# Example using a TS with frequency > 1 (in this case, frequency = 12)
tmpAir <- deseason(
  ts.data          = datasets::AirPassengers,
  ts.frequency     = stats::frequency(datasets::AirPassengers),
  forecast.horizon = 20
)

head(datasets::AirPassengers) # original data
head(tmpAir$deseasonTS)       # deseasonalized series w/o multiplicative
                              # component, not equal to the original
head(tmpAir$deseasonTS * tmpAir$seasonalComp) # this should be equal to the
                                              # original data
# Example using a TS with frequency = 1
tmpUsage <- deseason(
  ts.data          = datasets::WWWusage,
  ts.frequency     = stats::frequency(datasets::WWWusage),
  forecast.horizon = 10
)

head(datasets::WWWusage)  # original data
head(tmpUsage$deseasonTS) # since frequency = 1, results should be the same

gu-stat/gears documentation built on Oct. 20, 2021, 2:53 a.m.