I'm excited to share my latest data science project where I delved into the world of car prices using a dataset from CarDekho.com. This project involved extensive data cleaning, feature engineering, and applying various regression models to predict car prices. Here are some key insights and findings:

🚗 Exploring Car Prices with Linear Regression: A Data Science Project on CarDekho.com Dataset

🚗📊 Project Overview:Dataset: CarDekho.com vehicle datasetObjective: Predict car prices using linear regression and other regression techniques

Tools Used: Python, Pandas, Seaborn, Scikit-learn, Statsmodels

🔍 Key Insights:

Diverse Car Range:

The dataset includes a wide variety of cars, from budget-friendly models to luxury vehicles priced up to ₹3.5 crores.Majority of the cars are family cars with a seating capacity of 5.

Car Usage:

The mean kilometers driven is around 52,000, indicating that most cars are well-used before being sold.

Popular Makes and Models:Top Makes: Maruti Suzuki, Hyundai, Mercedes Benz, Honda, ToyotaFuel and Transmission Types:Fuel Types: Diesel (1049), Petrol (942), CNG (50), Electric (7)Transmission Types: Manual (1133), Automatic (926)Geographical Insights:

Top Locations: Mumbai, Delhi, Pune, Bangalore, Hyderabad

Cars are categorized into regions (North, South, East, West) for better analysis.

Model Performance:Linear Regression:

R² (Test): 0.8619

Mean Absolute Error (MAE): 0.3137

ElasticNet Regression:R² (Test): 0.8619

Mean Absolute Error (MAE): 0.3065

Best Alpha: 0.0046,

Best L1 Ratio: 1.0

Feature Importance:

Top Features Influencing Price:

Age of the car

Engine Power (BHP)

Category (Luxury, Mid Range)

Fuel Tank Capacity

Transmission Type (Manual/Automatic)

Interesting Finds:Costliest Car: Ferrari 488 GTB priced at ₹3.5 croresCheapest Car: Tata Nano Base priced at ₹49,000Most Driven Car: Renault Duster with 2,000,000 kmLeast Driven Car: MINI Cooper JCW Hatchback with 0 km

This project was a fantastic opportunity to apply data science techniques to real-world data and uncover meaningful insights

Feel free to check out the detailed analysis and code on my GitHub https://lnkd.in/gHbfbN7b.

Also available on Kaggle - https://www.kaggle.com/code/manubindra/price-predictions-for-car-dekho-com/edithttps://www.kaggle.com/code/manubindra/price-predictions-for-car-dekho-com/edit