Abstract—In order to survive in today telecommunication market it is essential to have the ability of distinguishing customers who are probable to switch into a competitor. Customer churn prediction is a means to address the complication and has become an important issue in Telecommunication business. In such competitive market a reliable means to predict customer future’s action would be regarded as priceless. To end, this paper has employed Meta-heuristic, Machine learning, Neural Network and data mining techniques including Genetic Algorithm, Particle Swarm Optimization, Support Vector Machine, Artificial Neural Networks, Decision Tree, and K-Nearest Neighbors so as to solve churn prediction problem. Using the data of an Iranian mobile company not only these techniques were experienced and were compared to each other, but also we drawn a parallel between some different prominent data mining software. However, the result indicates that this paper ANN performed the best with near 90 percent precision and recall.
Index Terms—Telecommunication, churn prediction, data mining, meta-heuristics, ANN, KNN, SVM, decision tree, GA, PSO.
Abbas Keramati is with University of Tehran, Iran (e-mail: keramati@ut.ac.ir).
Ruholla Jafari Marandi is with the Department of Industrial and Systems Engineering, Mississippi State University, United States.
[PDF]
Cite: Abbas Keramati and Ruholla Jafari Marandi, "Addressing Churn Prediction Problem with Meta-Heuristic, Machine Learning, Neural Network and Data Mining Techniques: A Case Study of a Telecommunication Company," International Journal of Future Computer and Communication vol. 4, no. 5, pp. 350-357, 2015.
