Material Selection for Electric Vehicles chassis using Machine Learning Naïve Baise Techniques

The selection of suitable materials for electric vehicle (EV) chassis structures is a critical engineering challenge that directly influences vehicle safety, structural integrity, weight reduction, and energy efficiency.Traditional material selection approaches rely on empirical methods and multi-criteria decision-making techniques, which may be limited in handling complex, high-dimensional datasets of material properties.In this study, a probabilistic machine learning framework based on Naïve Bayes classifiers is proposed for systematic selection of EV chassis materials.Key mechanical properties, ultimate tensile strength (Su), yield strength (Sy), Elastic modulus (E), shear modulus (G), Poisson's ratio (μ), and density (ρ), were used as input features to classify material suitability.Three models were evaluated: Gaussian Naïve Bayes (GNB), Kernel Naïve Bayes (KNB), and Optimizable Naïve Bayes (ONB).Among the implemented models, the comparative analysis revealed progressive performance improvements from GNB to KNB and then to ONB.The ONB model achieved the highest validation accuracy (91.98%) and test accuracy (88.39%), along with the lowest total misclassification cost, indicating enhanced predictive reliability and generalization capability.Receiver Operating Characteristic (ROC) curve analysis confirmed superior class separability of the optimized model.Feature importance evaluation demonstrated that strength-related parameters (Sy and Su) and density (ρ) are dominant factors in chassis material classification, consistent with strength-to-weight design principles in EV engineering.The results demonstrate that ONB provides an efficient, interpretable, and computationally scalable decision-support tool for EV chassis material selection.The proposed data-driven methodology reduces dependence on heuristic selection processes and offers a structured framework for multicriteria engineering optimization.This work contributes to integrating materials informatics into automotive structural design, supporting the development of lightweight, high-performance electric vehicles.