Regression Analysis for Predicting Wood Pulp Demand by PSO Optimization
In today’s world, consumption of paper and paperbased products is increasing in all the fields. Wood pulp which is extracted from the wood chips is the most commonly used raw material to manufacture the papers. Demand and supply of the wood pulp determines the social-economical development of a country. Many forecasting methods are used to predict the future demands of the wood pulp so that the supply chain management can be planned. In this paper, support vector regression analysis methods are used to predict the demands of wood pulp and Particle Swarm Optimization (PSO) algorithm is proposed to optimize the parameters of kernel functions. Regression models were created by using the data collected from TNPL. The parameters such as Mean Magnitude Relative Error (MMRE) and Median Magnitude Relative Error (MdMRE) are used for evaluating the results. Evaluated result shows that proposed SVM regression with PSO approach gave improved accuracy with significant decrease in MMRE and MdMRE
