KLASIFIKASI OPPORTUNITY MENGGUNAKAN ALGORITMA C4.5, C4.5 DAN NAÏVE BAYES BERBASIS PARTICLE SWARM OPTIMIZATION

Abstract

Predicting an opportunity whether to be successful (buy) or not (no), with the aim to increase selling for target achievement of marketing. Marketing is required to find a good opportunity so that it can be a prospect of sales in great value and the long term. In this research, the dataset is taken from a CRM application (Customer Relationship Management) at PT. XYZ, sales of telesales team in January - March 2016. From these results, the PSO-based C4.5 algorithm has the highest accuracy and AUC value. In this research comparative algorithms C4.5, C4.5 based PSO, and Naïve Bayes using Confusion Matrix testing methods and ROC curves. The highest accuracy value using PSO-based C4.5 algorithm is 80,90% with AUC value 0.833 is good classification, next is Naïve Bayes based PSO algorithm with accuracy value equal to 83,15% and value of AUC 0,894 is good classification, the last C4.5 algorithm the lowest accuracy value of 66.67% with AUC 0.592 is failure classification. From these results, the PSO-based C4.5 algorithm has the highest accuracy.