Matching buying and selling of skills: An optimal skill selection problem in an online market

Abstract

The central function of the job market is to match the available job vacancies with job candidates. For job candidates, it is critical for them to be equipped with the right skills for gaining competitive advantage. In this thesis, we obtain a dataset by scraping publicly available information from job postings for data science/analytics/engineering and similar positions on an online job marketplace. In the past few years, demand for those data-related jobs has been on the rise, and many job seekers change their career path to work in this area. For that purpose, it is important for them to understand the pattern of demand for skills in the labor market and to identify the best skills to acquire for maximizing the number of job vacancies they can apply for. We address these issues based on the real-life dataset. First, through exploratory data analyses, we examine the correlation between the size of a company and the types of its required skills, as well as the correlation between the salary level offered by a company and the rating it receives on the online job marketplace. Then, we develop a linear integer programming model to formulate the skill selection problem to maximize the number of jobs covered by the selected skills. We show that the problem is NP-hard, and then solve it using both the commercial solver CPLEX and greedy heuristics.