Cost-Oriented Candidate Screening Using Machine Learning Algorithms

Abstract

Choosing the right candidates for any kind of position, whether it is for academic studies or for a professional job, is not an easy task, since each candidate has multiple traits, which may impact her or his success probability in a different way. Furthermore, admitting inappropriate candidates and leaving out the right ones may incur significant costs to the screening organization. Therefore, such a candidate selection process requires a lot of time and resources. In this paper, we treat this task as a cost optimization problem and use machine learning methods to predict the most cost-effective number of candidates to admit, given a ranked list of all candidates and a cost function. This is a general problem, which applies to various domains, such as: job candidate screening, student admission, document retrieval, and diagnostic testing. We conduct comprehensive experiments on two real-world case studies that demonstrate the effectiveness of the proposed method in finding the optimal number of admitted candidates.