Título: Previsão IPCA utilizando Árvores de Regressão com Variáveis Selecionadas por Dynamic Time Warping
Autores: Figueiredo, Karla;Mattos, Daiane
Resumo:
This work has developed a methodology involving several Machine Learning and Data Mining techniques to forecast the Extended National Consumer Price Index (IPCA), which is used by the Central Bank of Brazil as the official measure of inflation in the country. The model used to select variables by Dynamic Time Warping (DTW) and performs the prediction of the singlestep IPCA value with an uncertainty margin taken as the prediction of the class of that index. The results are promising and encourage the continuity of the study, especially with regard to the uncertainty margin for the index forecast.
Palavras-chave:
Inflation;Forecast;IPCA;Data Mining;Tree Regression
Páginas: 12
Código DOI: 10.21528/CBIC2017-101
Artigo em pdf: cbic-paper-101.pdf
Arquivo BibTeX: cbic-paper-101.bib