Indonesia Conference Directory

In telemarketing activity, selecting the most potential customers are important because can reduce processing time and operational cost. Therefore, the ability to select the most likely buying customers are urgently needed. In this study, we propose a clear sequence in doing telemarketing activity based on the previous telemarketing data which applying data mining technique. We weight the importance of 16 customers characteristics through 45,211 observations from a Portuguese bank. Applying Random Forest algorithm along with Information Gain Ratio as a criterion and 10-fold Cross Validation, the model able to weight the importance of attributes and achieves 90.01 % accuracy in predicting telemarketing success. Furthermore, the rank of attribute importance was designed to be a guidance map in selecting potential targeted customers as a managerial implication.