Análise de séries temporais para previsão do faturamento mensal de uma empresa

Abstract

Within the current market, it can be said that the company that knows its numbers and that has a good strategic planning, becomes a company that is prepared for changes in the market, and therefore more likely to success. The objective of this study is to obtain the forecast for the monthly invoicing of a wholesale company, to identify characteristics inherent to the series under study, such as trend, seasonality and white noise by means of speci c tests, observe the presence and e ects of possible interventions, and adjust time series models that represent the data appropriately. The methodology used in this study is the identi cation and estimation of a model as de ned by Box and Jenkins, more speci cally, a seasonal model represented by SARIMA(p, d, q)(P,D,Q). If it is possible that several models are adjusted for the same data set, we intend to use hypothesis tests and model comparison techniques, such as the absolute mean percentage error. After de ning the appropriate model for the time series, the forecasts were realistic about the observed values of the series.