AutoPricer – Used Car Price Predictor

Project Objective: Build a predictive model to estimate used car prices based on various features, helping users determine fair values and improving pricing transparency in the market.

Technologies Used: The project included data preprocessing, exploratory analysis, feature engineering, and training of regression models such as Linear Regression, Decision Tree Regressor, Random Forest Regressor, and LightGBM.

Techniques Applied:

• Transformation of temporal and categorical features.
• Outlier detection and removal.
• Model evaluation using R², MAE, and RMSE.
• Model comparison and selection based on prediction accuracy and training time.

Results: The Random Forest model achieved the best balance between performance and accuracy with an RMSE of 0.0836. The LightGBM model was the fastest to train (6.41s), while Linear Regression was the least accurate with an RMSE of 0.1498.

Skills Applied: This project demonstrated proficiency in regression modeling, data visualization, model evaluation, and performance optimization.

Conclusions: Key factors in vehicle pricing included brand, model year, transmission type, and fuel type. The resulting model can be used by dealers or individual sellers to suggest competitive and fair market prices.