The used car sales service Rusty Bargain needs an application that allows customers to quickly estimate the market value of their car. This project trains and compares multiple Machine Learning models to predict prices based on technical specifications, trim versions, and price history.

• Predict the market price of a used car.
• Compare models using three key criteria:
  • Prediction quality (RMSE)
  • Training time
  • Prediction time
• Implement a Linear Regression model with manual Gradient Descent.

The car_data.csv file contains historical used car data with features such as:

• Vehicle type, brand, and model
• Registration year and mileage
• Fuel type, transmission, and horsepower
• Sale price

• Python 3
• pandas, numpy — data manipulation
• matplotlib — visualization
• scikit-learn — baseline models, preprocessing, metrics
• LightGBM — gradient boosting
• XGBoost — gradient boosting
• joblib — model persistence

rusty-bargain-gradient-autos/
├── rusty-gradient-val.ipynb   # Main notebook
├── car_data.csv               # Dataset
├── regressor.json             # Model configuration
└── README.md

Each model was evaluated using RMSE on the test set, while also measuring training and prediction time. Gradient boosting models (LightGBM, XGBoost) provide the best prediction quality, while linear regression serves as a baseline reference. Detailed results and comparison charts are available inside the notebook.