Abstract

Recruitment of the most appropriate employees and their retention are the immense challenges for the HR department of most of the industries. Every year IT companies recruit fresh graduates through their campus selection programs. Usually industries examine the skills of the candidate by conducting tests, group discussion and number of interviews. This process requires enormous amount of effort and investment. During each phase of the recruitment process, candidates are filtered based on some performance criteria. The problem domain is complex and the aspects of candidates that impact the recruitment process is not explicit. The intelligence of the recruitment process is spread among the domain experts and extracted through knowledge acquisition techniques. This research focuses on investigating the underlying criteria and tries to capitalize on the existing patterns, to minimize the effort made during the recruitment process. The approach here is to provide the insights through in-depth empirical characterization and evaluation of decision trees for the recruitment problem domain. Experiments were conducted with the data collected from an IT industry to support their hiring decisions. Pruned and unpruned trees were constructed using ID3, C4.5 and CART algorithms. It was observed that the performance of the C4.5 algorithm is high. The recruitment process differs for each industry based on the nature of the projects carried out. Experiments were conducted to determine the attributes that best fits the problem domain. Using the constructed decision trees discussions were made with the domain experts to deduce viable decision rules.

Authors

N. Sivaram¹, K. Ramar²
National Engineering College, Tamil Nadu, India¹, Sri Vidhya College of Engineering and Technology, Tamil Nadu, India²

Keywords

Recruitment Process, Decision Trees, Selection Criteria, Machine Learning