Mobile phone customers churn prediction by using Elman and Jordan recurrent neural network with reinforcement learning algorithm / Muhammad Syahir Mohd Ribuan

The number of mobile phone user increases consistently year by year. While gaining new customer is harder than maintaining existing one, various churn predictor engine has been developed to fulfill this purpose. Using Recurrent Neural Network in predicting churn is still new to this field. Same goes for Reinforcement Learning which is the Q-learning. For that reason, this project main purpose is to develop two famous Recurrent Neural Networks; Elman and Jordan, and also equipping them with QLearning; to predict the probabilities of mobile phone churning rates. The scope of this project is to evaluate the performance between ERNN and JRNN. This project is developed using Netbeans IDE and Java language. The final experimental result shows that JRNN able to give better accuracy prediction compared to JRNN.