SAVE THE DATE Tapia 2018 Orlando, FL September 19-22, 2018

2017 Tapia Conference

An Approach for Preference-based Matching in Residential University Roommate Markets


Presenter: Onyeka Emebo (Montclair State University)


Roommates are a fundamental aspect of college experience, and better college outcomes in terms of academic performance, social wellbeing and integration, psychological support etc. have been attributed to roommate compatibility. In an ideal roommate market, candidates’ preferences are a major factor in roommate matching. Howbeit matching in some roommate markets is usually done ad hoc based on either the candidate’s course or year of study, without considering the actual preferences of the candidates. Although many algorithms have been proposed to solve the stable roommate problem, this paper addresses two challenges of operationalizing the matching algorithm in the roommate markets of residential universities. These challenges includes: Impracticality of always obtaining stable matching, as in most practical cases, a sub-optimal stable matching will suffice; and impracticality of large number of candidates ranking all others in a strict order of preference. In this paper, we proposed a proximity heuristics and a roommate matching algorithm for college roommate markets for residential universities. Our algorithm utilizes Euclidean distance metric to measure similarities between preferences of the candidates to produce an ordered preference list, eliminating the need for candidates’ preference ranking; based on the list, a modified algorithm matches candidates and assigns each pair of roommates to a room. The utility of the proposed algorithm was demonstrated via an illustrative example, while the computational complexity was adjudged minimal compared to some existing algorithm. These preliminary results show that the proposed algorithm can be applied successfully in roommate markets of residential universities.