Automatic Orientation of Student Candidates Based on Memory-based Collaborative Filtering

Abstract

The traditional method of directing student candidates towards courses organized in Congolese universities and higher institutes, particularly at the UPN, has shown its limits. Indeed, it is generally based on the candidate's will, dictated by the opportunities offered by certain courses at the end of studies, the influence of friends or parents, and the ease of obtaining registration, rather than on the skills (abilities) already proven by the candidate in the humanities cycle. Consequently, we are witnessing recurring academic failures, the depopulation of certain departments, and a growing increase in the number of graduates in certain courses, which creates a large number of unemployed people in the country. This article proposes a machine learning algorithm for a recommendation system based on memory-based collaborative filtering. This algorithm allows to guide student candidates according to their skills already demonstrated in the humanities. To do this, two matrices were considered: the first, , of dimension , consisting of the averages of the ratings obtained by each candidate among candidates in each subject (course) among m optional subjects organized in the humanities, such as mathematics, physics, chemistry, French, Latin, accounting, etc. The second, the matrix , of dimension , consists of the weightings of each subject among m in each course among training courses organized in the different departments of universities and higher institutes. The product of these two matrices gives the product matrix of dimension , which represents the scores obtained by each candidate in each course . This matrix makes it possible to associate each candidate with each training course. To determine the score allowing the student candidate to be appropriately directed towards the course , the cosine similarity function was used. To realize this project, the Python programming language was used with the jupyter editor existing in the anaconda environment.