tsfgrnn (version 1.0.5)

grnn_forecasting: Time series forecasting using GRNN regression

Description

It applies GRNN regression to forecast the future values of a time series. The lags used as autoregressive variables are set with the lags parameter. If the user does not set the lags, these values are selected automatically.

Usage

grnn_forecasting(
  timeS,
  h,
  lags = NULL,
  sigma = "ROLLING",
  msas = c("recursive", "MIMO"),
  transform = c("additive", "multiplicative", "none")
)

Value

An object of class "grnnForecast". The function summary can be used to obtain or print a summary of the results. An object of class "gnnForecast" is a list containing at least the following components:

call

the matched call.

msas

the Multi-Step Ahead Strategy.

prediction

a time series with the forecast.

model

an object of class "grnnModel" with the GRNN model

Arguments

timeS

A numeric vector or time series of class ts.

h

A positive integer. Number of periods for forecasting.

lags

An integer vector in increasing order expressing the lags used as autoregressive variables. If NULL (the default) the lags are selected in a fast, heuristic way. It is also possible to use the values "FS" and "BE", in which case, the lags are selected using forward selection or backward elimination respectively. These techniques are feature selection approaches.

sigma

A positive real value or a character value. The smoothing parameter in GRNN regression. Two character values are possible, "ROLLING" (the default) and "FIXED", in which case the parameter is chosen using an optimization tool with rolling origin evaluation or fixed origin evaluation.

msas

A string indicating the Multiple-Step Ahead Strategy used when more than one value is predicted. It can be "MIMO" or "recursive" (the default).

transform

A character value indicating whether the training samples are transformed. If the time series has a trend it is recommended. By default is "additive" (additive transformation). It is also possible a multiplicative transformation or no transformation.

References

F. Martinez et al. (2022). "Strategies for time series forecasting with generalized regression neural networks", Neurocomputing, 491, pp. 509--521.

Examples

pred <- grnn_forecasting(USAccDeaths, h = 12, lags = 1:12)
plot(pred)