A New Algorithm of Onine Customer Classification Based on Ant Colony Algorithm

Effective and correct customer classification plays a key role in analyzing and estimating customer consumption patterns, establishing personalized marketing service system and differentiation management. A new customer classification algorithm is advanced based on ant colony algorithm. Firstly, based on consumer characteristics and behaviors analysis, an indicator system is designed which includes 21 indicators belongs to customer characteristics type variables and customer behaviors type variables respectively. Secondly, ant colony algorithm is improved through adjusting its transition probability dynamically to speed up model’s convergence and to build the selection strategy more conducive to the overall search, and then a new algorithm for customer classification is deduced and analyzed. Finally the experimental results verify not only the problem of convergence speed has been solved, but also the simplicity of the model structure and the accuracy of the classification are ensured when the new algorithm are used in electronic customer relationship management practically.