🚗 Car Price Prediction Machine Learning Model

Developed a machine learning model to accurately predict car ex-showroom prices based on 140 car attributes, providing valuable insights for informed decision-making in the automotive industry.

🌟 Project Overview

In today's rapidly evolving automotive market, determining the appropriate pricing for cars is crucial for striking a balance between affordability for customers and profitability for businesses. By analyzing various car features such as mileage, horsepower, fuel type, and more, this project aims to predict the ex-showroom price of cars.

Ex-showroom price refers to the base price of a car before any additional costs like taxes and registration fees are added. This model helps the automotive industry make data-driven decisions about car pricing, ensuring competitiveness and customer satisfaction.

⚙️ Goal

The objective of this project is to build a machine learning model that accurately predicts car prices based on various relevant features, enabling businesses and consumers to optimize their strategies in the competitive automotive market.

🎯 Purpose

This project provides valuable insights to businesses, helping car buyers and sellers make well-informed decisions. The machine learning model is built by analyzing a raw dataset with 140 car attributes.

📂 Dataset

• Source: The dataset was obtained from Kaggle, consisting of 140 car attributes.

🚀 Work Flow

1. Step 1: Importing the required libraries.
2. Step 2: Loading the dataset.
3. Step 3: Basic understanding of the data.
4. Step 4: Data Cleaning (handling missing values, outliers).
5. Step 5: Exploratory Data Analysis (EDA) and Insights.
6. Step 6: Data Preparation for Modeling.
7. Step 7: Model Building and Validation.

---

🔍 Methodologies and Techniques

• Feature Selection: Used SelectKBest based on statistical tests, such as ANOVA, to identify the most relevant features.
• Correlation Analysis: Performed to understand the relationships between features and target variables.
• Data Skewness: Checked and corrected using Box-Cox transformations.
• Scaling: Applied Robust Scaling to standardize the data.
• Model Comparison: Evaluated different machine learning models for performance.

🛠️ Technologies Used

• Programming Language: Python
• Libraries:
  • NumPy, Pandas (Data Processing)
  • Matplotlib, Seaborn (Data Visualization)
  • Scikit-learn, XGBoost (Modeling) and more.

⚡ Model Performance

Overall, the models showed excellent performance, with tree-based ensemble methods performing the best. These models captured complex relationships in the data, resulting in high accuracy.

Top 3 Models:

1. XGBoost Regressor:
   • 🏆 R² score: 98.7%
2. Random Forest Regressor:
   • 🥈 R² score: 98.4%
3. Gradient Boosting Regressor:
   • 🥉 R² score: 98.0%

---

📊 Key Insights

• XGBoost emerged as the top-performing model with an R² score of 98.7%, demonstrating its effectiveness in capturing complex interactions in the data.
• Tree-based models like Random Forest and Gradient Boosting also performed exceptionally well, with scores of 98.4% and 98.0% respectively.
• The dataset was cleaned, preprocessed, and reduced to the most significant features, enhancing model accuracy.

---

📈 Conclusion

This project successfully predicted car ex-showroom prices using relevant car attributes with a strong predictive performance of 98.7%. Key highlights include:

• Comprehensive Data Cleaning and Preprocessing steps to handle missing values, outliers, and inconsistencies.
• Effective Feature Selection techniques to identify the most important car attributes.
• Implementation of robust machine learning models, with XGBoost proving to be the best performer.

🙌 Acknowledgments Special thanks to Kaggle for providing the dataset and the community for inspiring this project.