Summary

• This project aims to create predictive models for car prices (MSRP) based on various car characteristics to help car manufacturers set optimal prices and improve growth. Three models were developed, analyzing different combinations of explanatory variables. The results show promising performance, with low MAPE, no overfitting, and significant explanatory power in each model

• Background: The models were built using machine learning algorithms and existing data on car characteristics to predict car prices, specifically Manufacturer Suggested Retail Price (MSRP) in the United States

• Models Used: Linear Regression (Ridge and Lasso)

• Conclusion: The final models performed well, with low error metrics and no overfitting. To enhance the models further, future steps may include feature selection using RFE or PCA analysis, creating separate models for luxury vehicles, and exploring other ML algorithms for comparison. Communicating the findings to business leadership in the automotive industry, it is evident that engine horsepower, engine fuel type, transmission type, driven wheels, vehicle size, and vehicle style significantly influence car prices (MSRP)