KOMPARASI 5 METODE ALGORITMA KLASIFIKASI DATA MINING PADA PREDIKSI KEBERHASILAN PEMASARAN PRODUK LAYANAN PERBANKAN

Abstract

Utilization data mining in banking marketing strategy is very effective. Prospective customer segmentation is one of the processes carried out in the banking marketing strategy. To support the results of the success rate of telemarketing personnel to market the product in its role of banking services that the process requires prospective customer data, then data mining support is very important in the classification of the prospective customers of the bank so that it can predict the degree of success in product marketing such services. Based on mapping studies of support data mining on prospective customers to come is no classification algorithms are often used for the classification of a borrower among others Neural Network, Naive Bayes, Decision Tree, K-NN and Logistic Regression, of this algorithm can result from the evaluation process by using Cross Validation, confusion matrix, ROC Curve and T-Test to determine the classification of data mining algorithms are the most accurate in predicting success in product marketing telemarketing services from the bank to do trials in the Neural Network algorithm was more accurate with an accuracy of 89.71% the AUC value of 0872, this may be a comparison of data mining classification Seeing AUC values of the five methods, then five groups of classification algorithms including both because of its AUC value between 0.80-1.00.