Today, with the extension of learning management systems (LMSs) and the diversity of learners' needs for online learning, instructors have to be assisted to adapt their syllabus to meet learners' needs. Therefore, it is necessary to tailor course instruction to meet individual needs and determine how well they serve the learners using these online platforms. In this case, technological advances are used to enhance e-learning by personalizing the learners' learning styles. For instance, gathering traces of systemic and contextual knowledge about learners and their learning preferences contribute to the design of a meaningful learning experience for learners. Our study, based on a questionnaire and learning traces, focuses on predicting learners' styles. The Felder Silverman Learning Style Model (FSLSM), among the best models in technology-enhanced learning, was applied to run an unsupervised clustering technique to cluster learners by preference degree in terms of profile and context for sequential/global dimension of the FSLSM. This paper presents the attributes of the learning contextual data-driven model which can be auto-populated and the appropriate data source determined to fill this model. To reach our aim, the data gathered from three agronomy courses taught in winter 2018, 2019, and 2020 in an LMS at the Hassan II Institute of Agronomy and Veterinary Medicine was analyzed. This paper concludes with the results achieved during the application of the proposed method in which most learners expressed their preferences as strong, balanced, or moderate for global and sequential learning styles in a predefined learning context.