Research On MTS Based On Rough Set Theory And Its Application In Customer Classification Of Direct Marketing In Banks

Foreign Banks with advanced marketing strategy have had a huge impact on our country’s commercial banks, domestic customers’ demands for products and services also become more diverse and personal, it’s urgent for domestic commercial banks to change the traditional marketing strategy. Direct marketing working as a new customer-oriented marketing pattern is helpful to improve the competitiveness of domestic banks. While the key to the effective implementation of direct marketing is to find out the right classification method to classify and target customers.Different from the common classification methods, the MTS method’s analysis is based on data without the data distribution assumption. MTS can select effective feature variables and achieve the purpose of real dimension reduction. MTS has been applied in many fields. To overcome the shortness of MTS, rough set theory is introduced to construct an improved MTS. Also the improved MTS will be used for classifying customers in banks’ direct marketing. The paper’s research includes two main parts as follows:(1) The theory research of MTS based on rough set theoryThe traditional MTS uses the orthogonal table and SNR to select effective variables, but some scholars proposed that this method has some shortness for selecting variables. Using orthogonal table and the signal-to-noise ratio is not necessarily the best way. The rough set theory is introduced to replace the orthogonal table and SNR for selecting variables, also MTS based on the rough set theory is constructed to improve the mahalanobis space and increase the classification accuracy.(2) The application research of MTS based on rough set theoryTo study customer classification in banks’ direct marketing, data related with direct marketing in a Portugal bank from UCI data set is used. MTS based on rough set theory and traditional MTS are both used to classify customers by analyzing data. The numbers of effective variables and the classification accuracy are compared. The conclusion is as follows: MTS based on rough set theory can reduce more ineffective variables as well as improve classification accuracy. It’s helpful to target customers and simplify data collection work in banks’direct marketing in China. Also MTS based on Rough set theory can be used in other classification problems.