This paper explores the optimization of learning in physics by applying Markov models, taking into account the learning styles defined by Kolb. We conducted a study with 115 university students to evaluate the influence of different teaching methods on the transitions between these learning styles. Using Markov models, we identified three learning states: initiation, understanding, and application. A transition matrix was developed to show the probabilities of moving from one state or style to another. The analysis was performed by machine learning algorithms. The results show that active methods, such as group projects and experiential laboratory activities, promote better understanding of concepts. This study highlights the importance of adapting teaching practices to students' learning styles, allowing for increased personalization to maximize student engagement and success. The results offer practical recommendations for teachers to improve the effectiveness of physics teaching.