In this paper, we present an approach for online course evaluation based on learners' behaviors during the learning process, where the course creator can monitor the quality status of their online courses based on learners' learning outcomes and then intervene to improve the success rate. For this purpose, a set of criteria has been developed. These criteria concern learners' cognitive, affective and academic engagement. The proposed approach was adopted by a system consisting of four parts: a learning management system, a LOG preprocessing manager, a quality assessment manager, and a visualization and intervention manager. A first test was carried out on a sample of 33 students from the Department of Economics at the University of Guelma (Algeria), where the results obtained were very encouraging and promising.