Forecasting Telecommunication Traffic Using the Enhanced Stepwise Projection Multiple Regression Method


Horng-Jinh Chang, Feng-Jenq Lin




In this paper , one new technological forecasting method, called the Enhanced Stepwise Projection Multiple Regression (ESPMR), is proposed for dealing with the traffic loads that are measurement data but not real values. It combines the adjusting and detecting concepts of SPA (The Sequential Projection Algorithm) with the estimating method of linear regression model, and considers the periodical factor into forecasts if the date series has the periodical traces. Two empirical studies on the forecasting for the International Telecommunication Traffic Loads of Taiwan and the Metropolitan Telecommunication Traffic Loads of Taipei illustrate good and stable results by executing the ESPMR system.