This project explores whether the resale value of a car can be accurately predicted using features like brand, age, mileage, fuel type, and more, while also uncovering which attributes contribute most to price depreciation.
- Can we predict the optimal selling price of a car?
- What factors influence resale value?
- Which brands hold their value best?
- Do automatic cars depreciate faster than manual ones?
- Does high mileage equal lower price?
Kaggle.com - rosaaestrada/Car Resale Value Prediction
- Data - Contains raw data, preprocessed data, train and test data
- Jupyter Notebook - The full source code along with explanations as a .ipynb file
- Python Code - The full source code along with explanations as a .py file
- Results - Summary Statistics, Visualizations, and Final Evaluation of the project
- Report - Summary slides of entire project
- Python= 3.12.3
- Pandas= 2.2.2
- NumPy= 1.26.4
- Matplotlib= 3.9.2
- Seaborn= 0.13.2
- Scikit-learn= 1.5.1
- XGBoost= 3.0.2
- Rows (Observation): 301, Columns (Features): 9
- Key Numerical features:
- Year: Range: 2003-2018; Median: 2014
- Selling Price: Range: 0.1-35; Mean: 4.66; Median: 3.60
- Present Price: Range: 0.32-92.6; Mean: 7.63; Median: 6.40
- Kilometers Driven: Range: 500-500,000; Mean: 36,947; Median: 32,000
- Owner Count: Most cars had 0 previous owners; Max: 3
.png?raw=true)
- 🔧 Present price, car age, year, and kms driven are the strongest predictors of resale value.
- 📉 Automatic cars depreciate faster than manual ones, losing more value on average.
- 🚗 High mileage has little predictive power on resale price in this dataset (correlation ≈ 0.03).
- 🏆 Random Forest outperformed Linear Regression and XGBoost, achieving the highest accuracy (R² = 0.72).
- 💰 Brands like Vitara Brezza and Bajaj Avenger retained over 90% of their original price, signaling strong demand and low depreciation.