An increase in undergraduate registered students in universities largely grown last years. However, the number of graduates remains low. The main cause of this issue is the evasion and / or retention in higher education courses. Techniques and methods that assist the identification of causes and / or patterns are important in this area. The purpose of this paper is to propose methods based on graph mining to identify bottlenecks retention in undergraduate programs. The methods can be used both on graphs representing undergraduate programs and on graphs representing students grade reports. To evaluate the proposed methods, real data from a Brazilian federal higher education institution were used.
Keywords: Graph Mining, Educational Mining, Undergraduate Curriculum, Undergraduate Historical Curriculum