Previsao de demanda: Uma aplicacao dos modelos Box-Jenkins na area de assistencia tecnica de computadores pessoais

Demand forecasting is an important tool to aid on the determination of necessary resources of a given company. In this paper, the Box-Jenkins methodology was applied to analyze historical data of a personal computer repair company and provide a forecast for the number of service calls. The company studied presents three segments of clients: contracts, warranty, and on-call. As each client has it own characteristics, in order to better represent tendency and seasonality behavior through the Box-Jenkins models, a specific forecasting model was developed for each segment. The choice of the optimum models were based into graphic analysis and statistical tests, which lead to the decision of adopting the AR(1) model to foresee the number of contract clients, the ARIMA(2,1,0) model for warranty clients and the SARIMA(0,1,0)(0,1,1)12 seasonal model for on-call clients.