Comprehensive Clusterwise Linear Regression for Pavement Management Systems

Abstract

A comprehensive mathematical program was formulated to determine simultaneously (1) an optimum number of pavement clusters, (2) cluster memberships of pavement samples, (3) cluster-specific significant explanatory variables, and (4) estimated regression coefficients for pavement performance models (PPMs). Simulated annealing coupled with all-subset regression was proposed to solve the mathematical programming. The proposed algorithm was capable of identifying and addressing potential multicollinearity issues. All possible combinations of the explanatory variables were examined to select the best model that provided a balance among (1) the number of PPMs; (2) the number of explanatory variables; (3) the resources required to develop, maintain, and use these models; and (4) the explanatory power. For the data set used in this research, six-cluster models were determined as part of the optimum solution. The predictive capabilities of the resultant models were investigated, and results showed that the models provided few prediction errors without any overfitting issues.