Skip to content

ricardobreis/Time-Series-PIB

Repository files navigation

Time-Series-PIB

A Brief Explanation

The purpose of this project is to find a model with a good fit with the Brazilian PIB (GDP) time serie, aiming to forecast september, october, november and december 2019. The time serie is a set of 56 data points, from january 2015 to august 2019, and it was partitioned into two parts, validation (a set of 48 data points) and test (a set of 8 data points) datasets, a proportion of 85% and 15%. The project was presented at Fundação Getulio Vargas (FGV) Time Series Analisys class.

Validation And Test Datasets

Validation And Test Datasets

Accuracy

Accuracy of Models

Best Model

Linear Trend With Seasonality

PIB - Time Series

Forecasting September, October, November and December 2019

Forecast

PIB Time Series Dataset

http://www.ipeadata.gov.br/Default.aspx

  • Frequency: Monthly from 1990.01 to 2019.08
  • Source: Banco Central do Brasil, Boletim, Seção Atividade Econômica (Bacen / Boletim / Ativ. Ec.)
  • Currency: R$ (Millions)
  • Comments: Quadro: Contas nacionais. Obs.: Produto Interno Bruto (PIB). Estimativa do Banco Central.
  • Updated on: 23/09/2019