Community Crime Analytics Dashboard using Machine Learning and Hotspot Detection

This paper outlines the design and features of a community-focused crime analytics dashboard. The system is designed to provide actionable safety insights to tourists and local citizens by leveraging machine learning and data visualization techniques. The platform analyzes historical crime data from sources like NCRB/Kaggle to visualize trends and distributions. A key feature is the integration of a simulated real-time crime feed via NewsAPI, which displays the latest news articles as live alerts with an associated confidence score. The predictive core of the dashboard uses an XGBoost model to generate a crime risk score or predict the most probable crime type for a user-specified location and time. All information is presented through an interactive Streamlit dashboard, which features mapbased hotspot visualizations using OpenStreetMap. The dashboard is designed with a simple, tabbed interface for ease of navigation, making complex data accessible to a general audience.