Orders prediction for small IT company

Abstract

Reliable methodology for service orders prediction can significantly improve the quality of business strategy. It is very important to identify the seasonal behavior in order data to correctly predict customer demand and make appropriate business decisions. There are several methods to model and forecast time series with seasonal pattern. This paper compares seasonal naive, Holt – Winters seasonal, SARIMA and neural networks methods in order to evaluate their performance in prediction of the future values of time series that consist of the monthly orders in a small IT company