Particle swarm optimization in regression analysis: A case study

Abstract

In this paper, we utilized particle swarm optimization algorithm to solve a regression analysis problem in dielectric relaxation field. The regression function is a nonlinear, constrained, and difficult problem which is solved by traditionally mathematical regression method. The regression process is formulated as a continuous, constrained, single objective problem, and each dimension is dependent in solution space. The object of optimization is to obtain the minimum sum of absolute difference values between observed data points and calculated data points by the regression function. Experimental results show that particle swarm optimization can obtain good performance on regression analysis problems. © 2013 Springer-Verlag Berlin Heidelberg.