Answers are grounded in this paper and cite their sources.
Demonstrates resource deployment effectiveness in traffic incidents, highlighting AI's role in decision-making.
GridSense: Explainable Ensemble Learning for Traffic Incident Severity Prediction and Resource Deployment GridSense is an explainable AI decision-support system developed for the Bangalore Traffic Police as part of Gridlock Hackathon 2.0 (Theme 2: Bangalore Traffic Intelligence). The system predicts traffic incident severity (Low, Medium, High) using a weighted ensemble of LightGBM, CatBoost, and XGBoost models trained on historical traffic incident data with domain-engineered features. To improve transparency, GridSense provides gain-based feature importance explanations, retrieves similar historical incidents using cosine similarity, and generates deployment recommendations for officer allocation, barricades, and diversion routes. The system also incorporates a feedback mechanism that enables verified field outcomes to be recorded for future model improvements. This repository contains the research preprint describing the methodology, feature engineering pipeline, ensemble design, explainability framework, evaluation, and deployment architecture.
No takes yet. Share an insight, caveat, or question.
Tripathi et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: