Bank Branches Site Selection in Competitive Condition Using Genetic Algorithm

Selection the optimum location of bank branches is one of the most important decisions in banking activity with significant effect on its performance. Bank branches site selection is a competitive facility location problem that a bank in competition with other banks to increase itself market Share in order to achieve maximum benefit attempt to establish one or more its branches of the best places in the competitive market. Positioning bank branches is NP-Hard problem that with the increasing number of demand points and branches, computational complexity of the problem increases exponentially. On the other hand, the search space of problem solutions also is widespread so using simple GIS analysis functions as well as certain techniques and heuristic algorithms is not resolved easily. Then to solve it is better be used meta heuristic algorithms in accordance with the terms of problem. The present research is to find suitable locations for new branches of bank "A" in the market to compete with existing branches of bank "B" that genetic algorithm is used to optimize the location of branches in four zone of the municipality seven region in the tehran. Process so that the required data s have been collected from sources and have been prepared with GIS software s to entry of algorithm. Genetic algorithm's objective function consist of two objective function. One is maximize the total market share obtained by bank A (Market Share), the other is minimizing the total market share lost in the existing branches of bank A because for the arrival of a new branch of the bank A in a competitive market (Cannibalization). That with initial population of 250, 500, 750 and 1000 each was run 100 times. The results showed that the best objective function value is equal. After that obtained optimal locations from algorithm, displayed by GIS softwares to help bank administrators for spatial analysis and decision-making. Finally, comparison the results of the genetic algorithm with weighted sum in GIS Represent of the two techniques were much adapted.