generate_forecast_table: generate_forecast_table function

In yForecasting/salesforecasting: Sales Forecasting

generate_forecast_table function

Description

Executes a list of forecast models for a predefined set of horizons

Usage

generate_forecast_table(
  data,
  models = c("ets"),
  start,
  freq = 365.25,
  h = 1,
  orimax = 1,
  train_length = NA,
  opt = rep(0, length(models)),
  export = FALSE
)

Arguments

data	A dataframe containing the sales data. Here the columns are individual time series, the different rows are different periods of the time series.
models	A vector with string notations of different models. A selection can be made from c("naive", "seasonal naive", "holt-winters", "ets", "auto.arima", "stlf", "tbats", "nnetar", "intermittent") or their abbreviation c("n","sn","hw","e","a", "st", "t", "nn", "i"). Default is c("ets").
start	A vector containing Year, Month and Day info in the format of c(YYYY, MM, DD).
freq	Frequency of the timeseries, numerical value.
h	The forecast horizon, how many periods ahead are the models forecasting.
orimax	Maximum number of rolling origins. Default value is 1. This parameter can be set to a different number if the goal is to simulate a forecasting setup over a rolling origin.
train_length	Number of train periods. When the default value NA is kept, this parameter is set to the length of the data timeseries - (h + orimax - 1).
opt	Optimise model: A numeric vector with options for each model that is called in the argument models. This is 0 by default for all models, but can be set to model-specific options.
export	TRUE if the table with forecasts should be exported to a csv file. Default is FALSE

Value

A data frame with the (yft) format.

Author(s)

Yves R. Sagaert

Examples

# Daily example
data <- round(data.frame(AA=rnorm(100,100,10),AB=rnorm(100,500,10),
BB=rnorm(100,1000,10)))
generate_forecast_table(data, models=c("ets"), start=c(2019,7,1),
freq=365.25, h=7, orimax=1, train_length=NA, export=FALSE)
generate_forecast_table(data, models=c("ets"), start=c(2019,7,1),
freq=365.25, h=3, orimax=2)

# Monthly example
data <- round(data.frame(MA=rnorm(30,3000,300),MB=rnorm(30,7000,400),
MC=rnorm(30,200,50)))
generate_forecast_table(data, models=c("n","sn"), start=c(2019,7),
freq=12, h=3)

# Intermittent demand
data <- matrix(abs(round(rnorm(240,0,0.5))),nrow=60,ncol=4,
dimnames=list(NULL,c("IA","IB","IC","ID")))
generate_forecast_table(data, models=c("i","i","i"), start=c(2019,7),
freq=12, h=3, opt=c(0,1,2))

yForecasting/salesforecasting documentation built on April 29, 2022, 7:21 p.m.