Course clustering based on the phylogenic tree of students’ grades in architectural degree

Abstract

A large-scale correlational analysis was conducted on the grades of the students who completed their studies in the 1994 curriculum in the Barcelona School of Architecture. The study included the passing grades of a total of 3910 students corresponding to all 47 compulsory subjects, as elective courses were excluded. The objective of the research was to explore whether courses with related content were associated with the corresponding grades at an aggregate level, and identify similarities and dissimilarities in their clustering. The correlation between the grades in all courses was computed pairwise, and used as a distance measure to quantify the grade of affinity between them. Using this distance metric, the results were processed using hierarchical cluster analysis producing a tree structure according to this similarity/dissimilarity measure. The clustering results were visualized in a phylogenic tree, identifying the optimal number of clusters and coloring the results accordingly. The visualization of this tree matched very closely the grouping of courses according to departmental divisions, indicating a strong association between the grades obtained and specific areas of knowledge. The lowest marks were also found to be concentrated in the first years and among the graphic and design subjects. These results reinforce previous studies that have investigated the relationship between academic performance during the degree course and the entry grade. The great discontinuity between them and the fact that PBL is characteristic of the degree course and less common in secondary school could be the reasons for these results.