Forecasting by a Modified Exponential Smoothing Method

摘要

Exponential smoothing, a forecasting procedure that has been used in systems engineering, educational psychology, and advertising research, is discussed and a modification presented that increases its accuracy. The modified method is applied illustratively to forecasting the number of independent health insurance plans and the number of persons served by them from data series of unknown acceleration. The forecasting error is compared to the error in forecasts obtained by other standard methods.