International Journal of Mathematics And Computer Research

Abstract

Time series analysis and forecasting process plays an important role in business, atmospheric studies, insurance companies, banking sectors etc. There are
many forecasting techniques like moving averages, double moving average, multiple moving averages, simple exponential smoothing, adaptive smoothing,
double exponential smoothing, triple exponential smoothing, autoregressive integrated moving averages, etc. In simple exponential smoothing, constant ‘α’ is
not fixed, it may vary from 0 to 1. In this paper, we discuss about ‘α’ where it is estimated through some process. We estimating the constant of exponential
smoothing using autoregressive moving average models by various autoregressive and moving average parameters. We fitting a new exponential smoothing
model by fixing value to ‘α’. To check for a goodness of fit, we use Kolmogrov - Smirnov test to simple exponential smoothing and new exponential smoothing
models. Mean square error criteria are used for the purpose of choosing best model between simple exponential smoothing model and new exponential smoothing model.