Machine learning based misconduct pattern detection and prevention system in sustainable communities (Campus hostels)

Campus hostels are environments where students live, interact with people and develop socially. However, campus hostels are also spaces where various forms of misconduct such as theft, drug use and many more can occur and such misconducts affects student safety and disrupt the overall well-being of students and traditional methods of predicting misconduct have proved to be less effective due to their inability to handle complex data patterns. This project focuses on the development of a machine learning-based system for detecting and preventing misconduct in campus hostels. The system made use of three machine learning algorithms, which are Linear Regression, Random Forest and XGBoost, to analyze past misconduct data and identify patterns for predictions. Data preprocessing techniques such as encoding and normalization were carried out to improve model performance and accuracy. The findings showed that XGBoost performed best with an R-Squared Score of over 93%, Mean Squared Error of 1.16 and Root Mean Squared Error of 1.08, indicating strong predictive capability. The system was interfaced into a web-based platform to allow administrators receive predictions. The system provides features like data visualizations, analytical dashboards, and report generation to help administrators in decision making. This project shows that machine learning can improve misconduct detection and promote safer hostel environment for students.