title | output |
---|---|
MAFS - Multiple Automatic Forecast Selection |
github_document |
Current version: 0.0.2
mafs
is basically a wrapper for the forecast package. Its main function is select_forecast
, which takes a time series object as an input, splits it into training and test sets, fits up to (currently) 18 forecast models into the training set, measure their accuracy against the test set, chooses the best model according to the error metric defined by the user and outputs the results of the models and the forecasted future values.
mafs
is still at a very early phase with lots of room for improvement. I kindly invite all forecast analysts and practitioners to test my package and contribute to it, either by reporting issues or by pull requests.
For the time being, the development version is available only at Github.
# install.packages("devtools")
devtools::install_github("sillasgonzaga/mafs")
You can run a simple test the package with the code below
library(mafs)
select_forecast(AirPassengers, test_size = 6, horizon = 12, error = "MAPE")
Shiny app demonstration of the models included in mafs
.