A novel approach in multi response optimization for correlated categorical data

The main purpose of this paper is the optimization of multiple categorical correlated responses. So, a heuristic approach and a log-linear model have been used to simultaneously estimate the responses of surface parameters. Parameter estimation has been performed with the aim of maximizing the amount of concordance. Concordance means that the joint probability of occurrence of dependent responses in each treatment is more than other probabilities in the same treatment. The second step of this research is the optimization of multi correlated responses for categorical data using some practical metaheuristic algorithms, including simulated annealing, Tabu search and the genetic algorithm. Using each meta-heuristic algorithm, the best controllable factors are selected to maximize the joint probability of success. Three simulated numerical examples with different sizes have been used to describe the proposed algorithms. Results show the superiority of joint success probability values in the Tabu search algorithm, compared to the other approaches.