PREDIKSI LOYALITAS PELANGGAN TELEKOMUNIKASI MENGGUNAKAN LOGISTIC REGRESSION DENGAN SELEKSI FITUR PARTICLE SWARM OPTIMIZATION

Stefanus Santosa
Fenilinas Adi Artanto

Jurnal Teknologi Informasi Cyberku
Vol 11 No 1 (2015): Jurnal Teknologi Informasi CyberKU Vol.11 no 1

For many companies, finding a reason to lose customers, measurement of customer loyalty and regain customers have become essential, including for telecommunication companies. The telecommunications company is one of the industry, where the customer really needs special attention because it is very influential in maintaining the stability of the company's revenue. The telecommunications industry has always faced the threat of financial loss resulting from customer loyalty. The customer who leaves the service is usually called churners. Find churners can help telecommunications companies in retaining customers and keep the company financially. This study used Logistic Regression algorithm with feature selection Particle Swarm Optimization to predict customer loyalty telecommunications. The test results obtained using ANN algorithm accuracy value amounted to 94.80%, and Logistic Regression Algorithm with Particle Swarm Optimization feature selection shows the value of accuracy of 97.65%, and the AUC value of 0.99, then the Logistic Regression algorithm with feature selection Particle Swarm Optimization can improve the accuracy of prediction telecommunications customer loyalty