Abstract
Here we propose a predictive model that identifies customers who switches their brand loyalty. Our basic idea is that enumerates emerging sequential patterns and classifies customers by "CAESP". In our experiments, we select a target brand and its competitive brand, and express their discount price and a gap between them by some categories. By doing this, it can be shown not only a brand selection, but also a condition of selecting the brand. In addition to this, we introduce some taxonomy for the item. From our computational results using a practical POS data, we show that our method has good performance for the data.
