Network Traffic VPN Classifier

This project aims to classify network traffic as VPN or Non-VPN using a Random Forest classifier. The model is trained on a dataset with various features extracted from network traffic and saved for future use.

Project Structure

• nt_vpn_classifier.ipynb: Jupyter notebook containing the code to train and evaluate the Random Forest model.
• network_vpn_classifier.pkl: Saved model file for the Random Forest classifier.
• README.md: Project documentation.

Prerequisites

Before running the project, ensure you have the following installed:

• Python 3.x
• Jupyter Notebook
• Required Python packages

Installation

1. Clone the Repository:

   git clone git@github.com:akaasula21/network-traffic-classification.git
   cd network-traffic-classification
   

2. Create a Virtual Environment (optional but recommended):

   python -m venv nt_classi_env
   source nt_classi_env/bin/activate  # On Windows, use `nt_classi_env\Scripts\activate`
   

3. Install Required Packages:

   pip install -r requirements.txt
   

Usage:

Training the Model

1. Open Jupyter Notebook:

   jupyter notebook
   

2. Run the Notebook:

   Open nt_vpn_classifier.ipynb in Jupyter Notebook. Execute the cells to load the dataset, preprocess the data, train the model, and evaluate its performance.

3. Save the Model:

   After training, the model is saved as network_vpn_classifier.pkl.

Using the Saved Model

To use the saved model for predictions, follow these steps:

1. Load the Model:

   import joblib
   model = joblib.load('network_vpn_classifier.pkl')
   

2. Make Predictions:

   Prepare your input data to match the features used during training. Use the model to make predictions:

   predictions = model.predict(input_data)
   

Project Details

Features used:

• Duration
• Total FIAT
• Total BIAT
• Min FIAT
• Min BIAT
• Max FIAT
• Max BIAT
• Mean FIAT
• Mean BIAT
• Flow Packets Per Second
• Std FIAT
• Min Active
• Mean Active
• Max Active
• Std Active
• Min Idle
• Mean Idle
• Max Idle

Model:

Random Forest Classifier

Evaluation Metrics:

Accuracy, Precision, Recall, F1-Score

Contributing

Contributions are welcome! Please fork the repository and submit a pull request for any improvements or additions.

License

This project is licensed under the MIT License.

Acknowledgements

Thanks to the dataset providers for making this project possible.
Special thanks to the Python and machine learning communities for their support and resources.