The objective of this project is to develop a machine learning model that accurately predicts the price of used cars based on various attributes, such as model, color, year, mileage, condition, and other relevant features.
The dataset contains the information about the used cars from one of Germany's largest car sales websites, AutoScout24. This scraped dataset contains a wide range of information about car offers, covering a cars manufactured from 1995 to 2023.
Link: https://www.kaggle.com/datasets/wspirat/germany-used-cars-dataset-2023
Used the XGBoost, LightGBM, and CatBoost Algorithm
Evaluate the models using RMSE and R² metrics.
| Model | R² | RMSE |
|---|---|---|
| LightGBM | 0.912955 | 3314.699804 |
| XGBoost | 0.916546 | 3245.617915 |
| CatBoost | 0.917652 | 3224.034970 |
The CatBoost model achieved the best performance with an RMSE of 3224.03 and an R² of 0.917652. The model can be used to predict the price of used cars based on their attributes.
- Create a API to deploy the Model