This study examines the impact of key socio-economic factors—education, work experience, gender, and age—on income, with a particular focus on capturing heterogeneity across the income distribution. A structured preprocessing framework is employed, including missing value treatment, categorical variable transformation, and outlier detection using the Interquartile Range (IQR) method. The dataset is divided into training (80%) and test (20%) subsets to assess model generalizability, and K-fold cross-validation is applied to evaluate predictive stability. Ordinary Least Squares (OLS) regression is initially used to estimate average effects, followed by diagnostic tests to assess assumptions such as normality, homoscedasticity, and the presence of influential observations. To address these limitations, Robust Regression is employed to obtain stable estimates in the presence of outliers, while Quantile Regression is implemented at the 25th, 50th, and 75th percentiles to capture distributional heterogeneity. The results show that OLS estimates are sensitive to extreme observations and assumption violations, whereas Robust Regression improves stability. Quantile Regression reveals that the returns to education increase across higher income quantiles, indicating significant heterogeneity. Model evaluation using test data and cross-validation suggests that OLS achieves the lowest RMSE, while alternative methods provide deeper insights into the income structure. Overall, the study highlights the limitations of mean-based approaches and demon-strates that combining robust and quantile-based methods offers a more comprehensive framework for analyzing income inequality.
