ABSTRACT
Security is a big deal in this era of information and communication technologies. Organizations spend a lot of time, money, and effort to secure their data and prevent unauthorized access or cyber-attack, and many people have built their careers in cybersecurity. In the early stage, the freshers do not know about the salary and how the salary varies based on the size of the company, the skill of the employee, and other factors. The relationship between the individual variables and their effect on the cybersecurity salary is analyzed in this research. Five machine learning (ML) regressors, namely Lasso, Random Forest, Decision Tree, Light Gradient Boosting, and Adaptive Boosting, with our proposed ensemble ML model named Blending Decision Tree, Ridge, Random Forest Regression (BDRR). The proposed ensemble BDRR shows superiority over other regressors to predict the cybersecurity salary. The analysis shows that full-time employees get more salary than part-time, contract, and freelance employees. Also, during the COVID-19 period, the salary went down to before the pandemic, and after the situation, the salary increased significantly. The salary of executive-level employees is much more than that of senior, entry-, and mid-level employees. The maximum salary goes to remote-level workers; most companies are medium-level. To show the stability of our proposed model, we evaluate the result in the 80:20, 70:30, and 50:50 training testing ratios. In most cases, our proposed model outperforms other regressors and shows 0.04052 MAE, 0.00355 MSE, 0.05954 RMSE, 0.02398 MPD, and 34.21954 SMAPE. To show the significance of our proposed ensemble model, we apply the Diebold-Mariano Test (DM test) with significant p-values. This article will give an idea to the freshers about the salary, company, and the needed skills to develop.
