A Decision Support Software on Bidding for Job Interviews in College Placement Offices

Many university placement offices employ a bidding system to allocate on-campus recruiter interview slots to students. Typically, a student is given (say) 700 points each week to bid on the firms visiting that week. Interview slots for each firm are assigned beginning with the highest bidder until all slots are filled. This paper describes the mathematical modeling behind a decision support system for helping students to bid in such a system. It has three components. The first component elicits a student's utilities of getting an interview with the various firms. The second component estimates the probability of getting an interview with a particular firm for a given bid amount. The final component considers our bidding problem as the maximization of a student's expected utility, which can be formulated as a nonlinear integer programming (IP) problem. It is shown that this IP problem can be transformed into a number of nonlinear programming problems without integer requirements, which can then be solved very rapidly to give on-line bidding recommendations to a large number of students.