reichlab/sarima-utils: SARIMA Models with Transformations and Seasonal Differencing

A set of wrapper functions around the forecast package to simplify estimation of and prediction from SARIMA models. Estimation and prediction routines are provided for a "default" sarima model specification that we have found through experimentation that works fairly well for several infectious disease data sets. In this specification, incidence data are first transformed to approximate normality, then seasonal differencing is performed, and the transformed data are passed to forecast::auto.arima for model selection and estimation. A function to recursively simulate "trajectories" of future incidence is provided, based on functionality in the forecast package.

install.packages("remotes")
remotes::install_github("reichlab/sarima-utils")