Forecasting Through Motif Discovered by Particle Swarm Optimization Algorithm

Abstract

Forecasting is an operational research technique used as a basis for management planning and decision making. Forecasting provides clues and reduce risk and uncertainties. This technique has its own indispensable importance in a wide range of zones including meteorology, energy economics and the study stock market etc. The main goal of this work is to predict the occurrence of outliers in time series, based on the discovery of motifs by using Particle Swarm Optimization (PSO) algorithm. Forecasting the future values of the time series are carried out by the discovered motifs. Such forecasted values are compared with its counterparts obtained by different kind of methods like Vector Auto Regressive Prediction model, native modal and motif based forecasting using Genetic Algorithms. The achieved results prove the potency of the proposed PSO algorithm.