PREDIKSI KEBERHASILAN TELEMARKETING BANK UNTUK MENCARI ALGORITMA DENGAN PERFORMA TERBAIK

Abstract

To find algorithms that have the best performance in predicting the success of telemarketing in banking courses researchers have conducted various material tests of several algorithms for data from the uci data set, and have as many as 17 attributes, some algorithms that have previously been tested in this study. to find the best performing algorithm using algorithm authors, among others, is to use an algorithm based on particle swarm optimization to optimize some attribute values ​​and to improve the accuracy of algorithms and higher data classification, and can produce even higher accuracy values. From the neural algorithm network (NN) based on PSO, the results are 91.80%, Support Vector Machine (SVM) to get an accuracy of 90.20%. Naif Bayes (NB) with an accuracy of 89.41%, and to use the Decision Tree (DT) algorithm with an accuracy of 90.93%. Then the PSO Neural Network based algorithm is clear resulting in higher accuracy than some algorithms tested with an accuracy of 91.80%. These results are classified as very good (very good classification).