MILEAGE PREDICTION - REGRESSION ANALYSIS

Objective

The project aims to predict city-cycle fuel consumption (miles per gallon) using regression analysis. The dataset contains a mix of categorical and continuous attributes, which influence fuel efficiency.

Data Source

The dataset is sourced from the YBI Foundation.

Dataset Information

The dataset consists of 398 entries and 9 attributes:

• mpg: Miles per gallon (target variable)
• cylinders: Number of engine cylinders
• displacement: Engine displacement
• horsepower: Engine power output
• weight: Vehicle weight
• acceleration: Acceleration capability
• model_year: Year of the vehicle model
• origin: Manufacturing origin
• name: Vehicle name/model

Data Preprocessing

• Checked for missing values and removed them
• Standardized numerical features using StandardScaler

Data Visualization

• Pairplot and regression plots were generated using Seaborn to visualize relationships between features

Model Training

• Features (X): ['displacement', 'horsepower', 'weight', 'acceleration']
• Target (y): mpg
• Train-test split: 70% training, 30% testing
• Model: Linear Regression

Model Performance

Metrics calculated on test data:

• Mean Absolute Error (MAE): 3.33
• Mean Absolute Percentage Error (MAPE): 14.7%
• R-squared Score: 70.3%

Dependencies

Install the required libraries using:

pip install pandas numpy matplotlib seaborn scikit-learn

Running the Project

1. Clone the repository:

git clone <repository-url>

2. Run the script:

python mileage_prediction.py

Conclusion

The Linear Regression model achieved an R-squared score of 70.3%, meaning it explains 70% of the variance in mileage prediction. Further improvements can be made by feature engineering, adding more attributes, or using advanced regression techniques.

License

This project is open-source under the MIT License.