Particle swarm optimization algorithm-based nonlinear LS-SVM model for modelling and forecasting time-series data

In this article, a novel Nonparametric, Nonlinear Least Squares Support Vector Machine (LS-SVM) methodology is thoroughly studied. The Particle Swarm Optimization (PSO), which is a very efficient population-based global stochastic optimization technique, is employed to estimate the hyper-parameters and time lag of Nonlinear LS-SVM model for time-series modelling. Relevant computer program is written in MATLAB function (m file). The MATLAB and STATISTICA software packages are used for carrying out data analysis. Subsequently, as an illustration, the methodology was applied to all-India annual rainfall time-series data. Superiority of this approach over ANN model is demonstrated using Root Mean Square Error (RMSE) and Mean Absolute Percent Error (MAPE) criteria for data under consideration.