📘

This repository provides the full Python implementation for the paper "Explainable Air Quality Management: Adaptive Federated XGBoost Enhanced by SHAP Analysis and Sequential Anomaly Detection." It presents a scalable, interpretable, and privacy-preserving framework for urban air quality prediction using distributed IoT sensor data.

Key components include:

• Federated Learning with Adaptive Aggregation: XGBoost models are trained locally at edge nodes, and aggregated using performance-aware weights to handle non-IID data.
• SHAP Explainability: SHapley Additive exPlanations are used to provide both global and instance-level feature importance for AQI predictions.
• PrefixSpan Anomaly Detection: A sequential pattern mining algorithm is applied to identify temporal irregularities in pollutant patterns.
• Real-world Deployment: Evaluated across 22 districts in Tehran with accuracy reaching 98.64%, MSE as low as 15.52, and explainability insights highlighting PM2.5, NO₂, and PM10 as key contributors.

📊 Dataset Access The dataset used in this study, titled “Anomaly-Aware Air Quality Monitoring with SHAP and PrefixSpan in Federated IoT Networks”, is publicly available on Kaggle: 🔗 https://www.kaggle.com/datasets/saghar001/air-quality-prediction-case-study

🔧 Code Repository The complete codebase is available on GitHub: 🔗 https://github.com/SagharShafaati/AdaptiveFL-XGB-QuantumXAI