Modeling of Time Series Data Prediction using Fruit Fly optimization Algorithm and Triple Exponential Smoothing

Abstract

This paper proposes a Prediction model, which optimizes the Triple Exponential Smoothing (TES) alpha, beta and gamma parameter algorithm using Fruit Fly optimization Algorithm (FOA) algorithm to predict time series data. Based on the research of previous authors, the TES algorithm is very likely to be sensitive to changes in the constants to the 3 parameters, where to do benchmarking it results use the MAPE method. Therefore, the authors limit this research by optimizing the parameters of the TES algorithm with the FOA algorithm. The dataset used in this experimental study has datasets which obtained publicly from the data market website repository. From this study, it was found that combination of Fruit Fly optimization Algorithm – Triple Exponential Smoothing (FOA-TES) can predict the time series data well with the average MAPE of 6%, better than the TES with an increased MAPE as 4%.