Surface quality of engineering components directly affects their performance and service life. Conventional machining processes, particularly under conditions that induce high surface roughness and tensile residual stresses, are often unable to ensure the desired final surface quality. Consequently, the application of surface enhancement techniques such as deep ball burnishing (DBB) is essential for improving surface integrity. In this study, the combined effects of key DBB process parameters, including ball diameter, penetration depth, and feed rate, on surface quality indices—Ra, Rz, and surface microhardness—were analyzed. Advanced machine learning models were employed for predictive modeling, with the SVR model demonstrating the best performance, achieving R² values of 0.925, 0.942, and 0.910 for Ra, Rz, and microhardness, respectively. Although the XGBoost model also produced acceptable predictions, its accuracy was lower than that of SVR. The use of partial dependence plots (PDPs) enabled a quantitative assessment of the relative influence of the input parameters, revealing that burnishing penetration depth was the dominant factor for all outputs, accounting for approximately 41–47% of the predicted variations in surface quality. Nevertheless, ball diameter and feed rate also exhibited substantial contributions, each ranging from approximately 23–30%, highlighting their non-negligible roles in DBB process control and optimization. Overall, this study presents a novel machine-learning-based framework for analyzing and optimizing the DBB process to enhance surface quality.