Mo-Resa77/Network-Anomaly-Detection-KMeans

# Network Anomaly Detection using Unsupervised KMeans Clustering An advanced, unsupervised machine learning framework and continuous behavioral audit pipeline engineered to analyze network logs and detect security intrusions based on live performance infrastructure attributes (Latency and Throughput). ## Core Architecture & Engineering Features - **Unsupervised Spatial Modeling:** Automated framework trained completely blind to target ground-truth vectors to establish an authentic network baseline topology. - **Traffic Feature Analysis:** Processes continuous architectural feature metrics—specifically mapping Network Latency (ms) against Data Throughput (Mbps) metrics. - **Data Balance Optimization:** Built-in programmatic cross-validation accounting for severe class imbalances (393 benign system entries vs. 12 sparse threat records). - **Statistical Metric Cross-Tabulation:** Leverages Pandas cross-tabulation (Confusion Matrix alignment check), Recall optimization, and precision evaluation to capture malicious traffic boundaries. ## 🛠️ Technology Stack - **Languages:** Python - **Machine Learning Library:** Scikit-Learn (sklearn) - **Data Visualizations:** Seaborn, Matplotlib - **Data Analytics Engine:** Pandas, NumPy - **Environment:** Google Colab / Jupyter Notebooks ## Core Engineering Insight Exposed Video link = https://drive.google.com/drive/folders/1-Om2MQpr5XZNH6l-J41VjAhyZ8fTkheH?usp=drive_link video of my work. Thanks.