Comparison of pattern matching Fuzzy Models and Box- Jenkins models To predict suscriptions case : Mobilis (2009-2012)

Abstract

This study is designed to forecast Subscriptions service prepaid in Mobilis, by approach of Box-Jenkins and pattern matching Fuzzy Models and to choose the best model of family models ARIMA.

The research found that the model box-Jenkins has achieved the highest forecast accuracy by comparing pattern matching Fuzzy Models, and the time series of subscriptions for the service prepaid in Mobilis is non-stationary, and to make it stationary the first differences are applied.

The best model among the models that have been developed in this study for forecasting the number of subscriptions is an ARIMA model (2.1.0).