A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Information Management, School of Engineering and Technology

Thesis (M.Eng.) - Asian Institute of Technology, 2007

Today, there are intensive competitions in marketing world. Marketers have to proactively run their businesses. Customer targeting is an important marketing decision that companies should pay attention to. Using information systems to support decision-making to shortlist potential buyers for certain products based on previous customer profiles or customer's intrinsic values is one way in targeting customer. Therefore, because of the dynamic behavior of customers, there is a need to respond to such changes immediately. Artificial immune recognition system (AIRS) classifier using value difference metric (VDM) as a distance metric has been created to predict whether customer will buy an insurance policy. Adaptive algorithm was proposed in order to respond to such dynamic change of customers. It can partially retrain or update classifier without starting training from scratch. Its performance was compared to the original training and cascade-correlation network. Results show that it can work well on continuous attribute type. Furthermore, fractal dimension reduction and Grassberger-Procaccia's algorithms were used together for attribute selection to gain predictive power of classifier. Moreover, difference between buyer and non-buyer in each attribute of profiles was also visualized to provide better understanding for marketers and to support marketing strategy in the future