Nowadays, e-learning service systems generate huge amounts of data, including both information directly about student behavioral and activity data within the information system during the learning process. System users’ activity and behavioral patterns during the learning process have a crucial role in determining learners’ needs and offer them the most suitable, appropriate learning content and personalized learning path, thus contributing to learning efficiency. Moreover, adaptive learning management systems ought to ensure an availability to identify possible gaps in the knowledge space of learners. In this chapter authors propose a new adaptive e-learning service system architecture, implementing machine learning technique and utilizing algorithms for predictive modeling, simulation, and forecasting, which allows the learning management system to learn from user activity data and offer them the unique personalized learning.