Abstract
It is needed to sustain hot-selling products for company which develops original products. To do so, it is important to grasp customer's reactions maintaining their own product philosophy, although customer's behaviors are so complicated. In this paper, we apply our proposed predict model to an actual business data described above to reveal the customer's reactions. Features in our method are that uses necessary and sufficient contrast patterns, and that shows highly readable relationship among the contrast patterns. By using the method, we reveal purchase behaviors of loyal customers in the company. From several computational experiments, we show that our method is practical one.
