Workplace Preference Analytics Among Graduates

Abstract

Graduates often find themselves difficult to secure a job after completing their education at universities or colleges. In this light, researchers have proposed various solutions to address this challenge. However, most of the work has largely focused on academic profile and personality traits; very few have highlighted the importance of workplace location characteristics. To address this challenge, this study has employed feature selection and machine learning approach to help graduates identify desired company type and sector based on their preferences and preferred location. The data used in this study was obtained from the Ministry of Higher Education Graduates Tracer Study's data, specifically for 2382 Multimedia University (MMU) students' employment situation upon graduating. Additional analytical datasets focusing on company and graduate locations were developed in order to extract further features relevant for this analysis. Feature selection was used to identify top-10 predictors that influence the selection of jobs in graduates' desired sectors. Various analytics methods such as Decision Tree Analysis, Random Forest Model selection, Naive Bayes Classification Method, Support Vector Machines and K-Nearest Neighbor Algorithms were employed for comparative evaluations within the workplace analytics scope. Notably so, results from this study demonstrate that using Random Forest Algorithm resulted in better performance in predicting employment status with an accuracy rate of 99.40%, predicting company type with 66.60% and lastly predicting company sector with 30.80% when compared to other predictive models utilized during our research work's project lifecycle phase.

Keywords

• classification
• machine learning classifier
• data visualization
• feature selection
• workplace preference analytics