Reducing Contention in an Undergraduate Capstone Project Allocation System

An online project bidding and allocation system used by the School of Electrical and Information Engineering, at the University of the Witwatersrand, is described and critically analysed. An important question that arose when designing the bidding system for projects was how to deal with the contention for highly popular projects. An approach, common in the literature, is to optimise the project allocation; however, the system implemented aims to incentivise students to consider less popular projects. Projects are allocated to groups using a quasi-random algorithm, as opposed to an optimal algorithm. Over the eleven year analysis period the percentage of first choices awarded has ranged from 57.5% to 80.8%, while the average number of unallocated groups per year is 8.2%. In addition to these outcomes project contention is shown to reduce during bidding window period in nine out of the eleven years of the study. This is due to novel aspects of this system including the transparency of the allocation algorithm and the fact that groups are able to see all competing bids for the projects they are bidding on and change their own bids in response. The allocation algorithm is, by necessity, sub-optimal in order to achieve the goals of transparency and fairness in that the stated winning probability, or a greater probability, for a bid always holds true when a project is allocated.