The Self-organizing Map (SOM) is one of the learning models widely used in market segmentation, and Growing Hierarchical SOM (GHSOM), which is a model extended to a hierarchical structure, is also used for the task. However, GHSOM cannot increase the map size due to the limitation of the number of data allocated to the underlying map. To aim for visual understanding of market data, we newly propose construction of a model through interacting with GHSOM analysts. In the analysis, we extract the newly defined indexes that show the customers behavior from the dataset as the feature vectors. Furthermore, market segments hidden in data set are visualized based on the method we propose.
