IntroductionThe telecom sector is witnessing a massive increase in data, and by analyzing this massive data, telecom operators can manage and retain customers. It is also important for companies to be able to predict the amount of income they may receive from their active customers. For this purpose, they need models able to determine customer loyalty. The cost associated with customer gain is usually higher than the cost associated Abstract Given the growing importance of customer behavior in the business market nowadays, telecom operators focus not only on customer profitability to increase market share but also on highly loyal customers as well as customers who are churn. The emergence of big data concepts introduced a new wave of Customer Relationship Management (CRM) strategies. Big data analysis helps to describe customer's behavior, understand their habits, develop appropriate marketing plans for organizations to identify sales transactions and build a long-term loyalty relationship. This paper provides a methodology for telecom companies to target different-value customers by appropriate offers and services. This methodology was implemented and tested using a dataset that contains about 127 million records for training and testing supplied by Syriatel corporation. Firstly, customers were segmented based on the new approach (Time-frequencymonetary) TFM (TFM where: Time (T): total of calls duration and Internet sessions in a certain period of time. Frequency (F): use services frequently within a certain period. Monetary (M): The money spent during a certain period.) and the level of loyalty was defined for each segment or group. Secondly, The loyalty level descriptors were taken as categories, choosing the best behavioral features for customers, their demographic information such as age, gender, and the services they share. Thirdly, Several classification algorithms were applied based on the descriptors and the chosen features to build different predictive models that were used to classify new users by loyalty. Finally, those models were evaluated based on several criteria and derive the rules of loyalty prediction. After that by analyzing these rules, the loyalty reasons at each level were discovered to target them the most appropriate offers and services.