SARIMA Analysis and Automated Model Reports with BETS, an R Package

Authors

Talitha F. Speranza, Instituto Brasileiro de Economia (IBRE)
Pedro C. Ferreira, Instituto Brasileiro de Economia (IBRE)
Jonatha A. da Costa, Instituto Brasileiro de Economia (IBRE)

Date of this Version

12-2018

Document Type

Article

Citation

The R Journal (December 2018) 10(2); Editor: John Verzani

December

Comments

Copyright 2018, The R Foundation. Open access material. License: CC BY 4.0 International

Abstract

This article aims to demonstrate how the powerful features of the R package BETS can be applied to SARIMA time series analysis. BETS provides not only thousands of Brazilian economic time series from different institutions, but also a range of analytical tools, and educational resources. In particular, BETS is capable of generating automated model reports for any given time series. These reports rely on a single function call and are able to build three types of models (SARIMA being one of them). The functions need few inputs and output rich content. The output varies according to the inputs and usually consists of a summary of the series properties, step-by-step explanations on how the model was developed, predictions made by the model, and a file containing these predictions. This work focuses on this feature and several other BETS functions that are designed to help in modeling time series. We present them in a thorough case study: the SARIMA approach to model and forecast the Brazilian production of intermediate goods index series.