Machine Learning-Driven Cost Forecasting in Automotive Production Engineering: Towards Sustainable and Intelligent Manufacturing

Abstract. Accurate forecasting of manufacturing costs remains a central challenge for the automotive sector, where volatile supply chains, dynamic energy prices, and sustainability pressures continuously reshape production systems. This paper proposes a machine-learning-based framework for cost prediction in automotive production engineering, integrating historical pricing data, supplier performance, and material indices. Several regression models were trained and compared, including Linear Regression, Decision Tree, Random Forest, Gradient Boosting, and XGBoost. The XGBoost algorithm achieved the best performance (MAE = 34.49, R² = 0.90), confirming its robustness and interpretability for complex industrial datasets. Furthermore, the study explores stochastic and hybrid modeling techniques, combining deterministic ML prediction with uncertainty quantification to improve reliability under variable market conditions. The proposed approach demonstrates the potential of machine learning to support sustainable and intelligent manufacturing, enhancing both economic efficiency and resource responsibility.