Customer churn classification in telecommunication company using rough set theory

Churn is perceived as the behaviour of a customer to leave or to terminate a service. This behaviour causes the loss of profit to companies because acquiring new customer incurred high investment for advertisements and promotions compared to retaining existing ones. Thus, it is necessary to consider an efficient classification model to reduce the rate of churn. In the traditional approach of classification modelling, it do not produce straightforward result interpretation. Therefore, identifying the best classification model to reduce the rate of churn is indeed a challenging task. The main objective of this thesis is to propose a new classification model based on the Rough Set Theory to classify customer churn. This research utilized the Knowledge Discovery in Database (KDD) process involving data pre-processing, data discretization, attribute reduction, rule generation, classification process, as well as data analysis, using the Rough Set toolkit. The Rough Set theory elements consist of indiscernibility relation, lower and upper approximations, as well as reduction set. Those elements are applied to classify customer chum from uncertain and imprecise dataset. The results of the proposed model are compared with a few established existing approaches. The results of the study show that the proposed classification model outperformed the existing models and contributes to significant accuracy improvement. The model is tested using dataset form local telecommunication company which achieves 90.32%. In conclusion, the results proved that the classification model based on Rough Set Theory had been capable to classify customer chum compared to the existing model