MODELING AND PREDICTING EMPLOYEE SALARIES USING MULTIPLE LINEAR REGRESSION

This study investigates the determinants of employee salaries using a multiple linear re- gression (MLR) model formulated through matrix algebra. Drawing on a dataset that includes education years, work experience, age, and weekly work hours, the research models salary levels in Nigerian Naira. Descriptive statistics reveal a professionally expe- rienced and well-educated workforce with considerable income variation. The regression model explains approximately 96.6% of the variance in salaries, with education and expe- rience emerging as the most influential predictors. Parameter estimates obtained via the matrix method align closely with results from the traditional lm() function in R, confirm- ing the model’s consistency and robustness. All coefficients are statistically significant, and variance inflation factor (VIF) diagnostics show no evidence of multicollinearity. Standard regression diagnostics support the assumptions of linearity, homoscedasticity, normality of residuals, and the absence of influential outliers. The findings underscore the predictive value of human capital and labor input factors in salary modeling. The matrix-based MLR approach proves to be both computationally efficient and statistically reliable, making it a valuable tool for compensation analysis, workforce planning, and policy formulation.