How chinmayee421/network_intrusion_detection_ml works
Project type and purpose
• This is a machine-learning-powered web application (full-stack ML app) for Network Intrusion Detection (NIDS).
• It ingests user-provided network traffic features via a Streamlit UI, applies a pre-trained Random Forest model, and returns “normal” vs. “malicious” predictions.
• The offline training pipeline (EDA, preprocessing, feature selection with RFE, model training in a Jupyter notebook) produces a serialized model artifact (random_forest_model.pkl).
File structure
• Top level:
• .github/workflows/main_team6.yml (CI/CD pipeline)
• PROJECT_TEAM_6_FINAL.ipynb (training, EDA, model development)
• nids.py (serving code that loads the model and makes predictions)
• random_forest_model.pkl (serialized model)
• requirements.txt
• README.md
• No separate “src” or “frontend” folder because Streamlit code lives alongside the serve script.
README for additional insights
• Tech stack: Python, Pandas/NumPy, Scikit-learn, Matplotlib/Seaborn, Streamlit UI, Azure Web App, GitHub Actions.
• Deployment targets: Azure Web App Services for hosting; GitHub for version control and CI/CD.
Putting it all together
• At the center: Azure Web App Hosting
• Contains a container or environment running Streamlit + nids.py
• On the left: End User (Browser)
• Sends HTTP requests to the Streamlit front-end
• On the right: Model artifact store
• random_forest_model.pkl loaded by nids.py
• Above or below: CI/CD pipeline
• GitHub → GitHub Actions → Azure Web App
• Optionally, show the offline training pipeline (Jupyter notebook → Pickle) as a separate branch