Hybrid Datamining Approaches to Predict Success of Bank Telemarketing

Telemarketing is a kind of straightforward marketing in which salesman requests the consumer either face to face or telephone request and influence him to purchase the product. Telemarketing achieves most prevalence in the 20th century and still increasing it. Now, the phone has been broadly accepted. It is valued efficient and holds the consumers up to date. In the Banking area, marketing is the backbone to exchange its goods or service. Business promotion and marketing is frequently based on an exhaustive understanding of actual information about the market and the real client demands for the productive bank manner. We recommend a data mining (DM) method to foretell the achievement of telemarketing requests for contracting long-term bank deposits. A local Portuguese bank was labeled, with data gathered from 2011 to 2016, thus involving the effects of the current economic crisis. We examined a comprehensive set of 11 features associated with bank consumer, goods and social-economic characteristics. We also discuss four DM forms with the hybrid model: Naïve Bayes (NB), Decision Trees (DTs), Perceptron Neural Network (NN) and Support Vector Machine (SVM). The four types were tested and compared with proposed hybrid classification methods (Perceptron Neural Network + Decision Tree) on an evaluation set, and we are splitting data into training and testing sets using cross-validation method. The proposed hybrid classification technique presented the best results (Precision 99% and ROC = 97%).