telescope.forecast: Perform the Forecast

In DescartesResearch/telescope: Hybrid Multi-Step-Ahead Forecasting

Description

Forecasts a given univariate time series in a hybrid manner and based on time series decomposition

Usage

telescope.forecast(
  tvp,
  horizon,
  rec_model = NULL,
  natural = TRUE,
  boxcox = TRUE,
  doAnomDet = FALSE,
  replace.zeros = TRUE,
  use.indicators = TRUE,
  save_fc = FALSE,
  csv.path = "",
  csv.name = "Telescope",
  debug = FALSE,
  plot = TRUE
)

Arguments

tvp	The time value pair: either vector of raw values or n-by-2 matrix (raw values in second column), or time series
horizon	The number of values that should be forecast
rec_model	Optional parameter:
natural	Optional parameter: A flag indicating wheter only natural frequencies (e.g., daily, hourly, ...) or all found frequencies shall be considered.
boxcox	Optional parameter: A flag indicating if the Box-Cox transofrmation should be performed. It is not recommend to disable the transformation. TRUE by default.
doAnomDet	Optional parameter: Boolean whether anomaly detection shall be used. FALSE by default
replace.zeros	Optional parameter: If TRUE, all zeros will be replaced by the mean of the non-zero neighbors. TRUE by default
use.indicators	Optional parameter: If TRUE, additional information (e.g. a flag wheter there is a high remainder) will be returned. TRUE by default
save_fc	Optional parameter: Boolean wheter the forecast shall be saved as csv. FALSE by default
csv.path	Optional parameter: The path for the saved csv-file. The current workspace by default.
csv.name	Optional parameter: The name of the saved csvfile. Telescope by default.
debug	Optional parameter: If TRUE, debugging information will be displayed. FALSE by default

Value

The forecast of the input data

Author(s)

Andre Bauer, Marwin Zuefle

Examples

telescope.forecast(forecast::taylor, horizon=10)

DescartesResearch/telescope documentation built on Oct. 23, 2021, 9:51 a.m.