A Time Series Forecasting Method Using Adaptive Exponential Smoothing Moving Average

Abstract

Time series forecasting is an essential issue across various fields, particularly for capturing demand fluctuations in sectors like business, economics, network, and inventory management. To address this issue, we propose AESMA (Adaptive Exponential Smoothing Moving Average), a method that combines exponential smoothing and simple moving average techniques to adaptively respond to recent data changes. By placing emphasis on recent observations while accounting for historical trends, AESMA effectively balances short-term fluctuations with long-term patterns. This adaptive capability enhances forecasting accuracy, making AESMA highly applicable for datasets with sudden demand shifts and seasonal fluctuations.