In traditional education, instructors enjoy the opportunity to observe student behaviors, which is a key contribution to testing and evaluation. On the other hand, observation is impossible in distance education. The purpose of this study is to offer a solution to the problem. In distance education, student performance can be tracked thanks to logs in the learning management system. These logs enable one to record how long a student studies teaching materials, how long he/she is active in the system, how successful he/she is in the quizzes taken, how active he/she is in the forums on the subjects, and how many messages he/she posts or reads. An analysis of these logs allows student performance to be predicted in the middle of the semester.