A New Robust Classification Method for CRM

Customer classification is a key step in customer relationship management (CRM), and there are many methods used for it, such as Neural Net, association rules, SOM model, etc. However, most existing methods don´t take noise which is very common in reality into consideration. In this paper, we combine Croup Method of Data Handling (GMDH) with Takagi and Sugeno fuzzy model (TS) to form a new classification method TS-GMDH. The experimental result shows that TS-GMDH outperforms the benchmark classifiers when the noise level is high.