PREDICTIVE ANALYTICS FOR USED CAR PRICING USING R AND REGRESSION METHODS

This study investigates the pricing dynamics of used cars using Multiple Linear Regression (MLR) and Ridge Regression techniques in R. The MLR model achieved a strong explanatory performance (R² = 0.8961), identifying key variables such as brand category, car age, and geographic region as significant predictors of price. However, issues like multicollinearity and overfitting limited its robustness. To address these challenges, Ridge Regression was employed, incorporating regularization to stabilize coefficient estimates and enhance predictive accuracy. Optimal lambda selection through cross-validation further improved model generalizability. The Ridge model confirmed key market trends, including depreciation with car age and premium valuation of electric and luxury vehicles. This dual-model approach not only demonstrates the comparative strengths of regression techniques but also provides actionable insights for stakeholders, offering a data-driven foundation for pricing strategy and policy development in the used car market.