sabafraz/network_anomaly_detection_machine_learning repository overview

⁠Flask Application

1. Pull the Docker image

docker pull sabafraz/network_anomaly_detection_machine_learning:1.0

2. Run the Flask app

docker run -p 5000:5000 sabafraz/network_anomaly_detection_machine_learning:1.0

⁠Network Anomaly Detection Machine Learning Pipeline On GitHub⁠

This project focuses on detecting anomalies and malicious activities in network traffic using Machine Learning techniques.The model can classify network traffic as normal or attack-related.

⁠Dataset Sources

Network traffic data 59410 samples from UNSW_NB15⁠

⁠Key Technical Components

1. Data Preprocessing Handled missing and inconsistent values

Converted boolean values to numerical format

Applied encoding techniques for categorical features:

Target Encoding to 'Proto' column

One-Hot Encoding to Other Categorical Columns

Applied scaling/normalization for numerical stability

2. Machine Learning Models

The project implements and compares multiple ensemble models:

Logistic Regression

Random Forest

Gradient Boosting

XGBoost

Gradient Boosting

3. Model Evaluation

Evaluated model performance using:

Accuracy

Confusion Matrix

Classification Report

Verified correct handling of unseen or missing features

⁠Performance Results

Best Model: Random Forest

Accuracy: 1.0

ROC-AUC: 1.0

Random Forest - Confusion Matrix:

[[10262     0]
 [    0  7561]]

⁠👨‍💻 Author

Developed by Saba Faraz
📧 Email: [email protected]⁠