Student feedback is one of the most relevant sources to evaluate the quality of the course and make improvements based on the comments students make about the classes. It is demonstrated that the opinion of the students influences the constant improvement of the instructors that teach the courses. Usually, these surveys contain Likert scale questions and open ended questions. However, academic coordinators usually focus on the quantitative answers and rarely read the open answers. To solve this issue, the current study analyzes the student feedback and synthesizes the data using Natural Processing Language (NLP) techniques such as topic modeling and emotion analysis, to diagnosticate and evaluate the point of view of the students. The source of this investigation is students' feedback from courses from a Colombian university located in Bogota in the first period of 2022. Additionally, using CRISP DM methodology was used to approach the solution to obtain adaptive models to the context of students surveys. This investigation shows that using models such as Latent Dirichlet Allocation(LDA) can synthesizes text information to obtain clústeres of words to complement and obtain different aspects that were not measure in the quantitative answers such as the ability to relate with students in different ways rather than being only teacher and student or the showing and communicating the required knowledge to dictated a course.