INTEGRATION OF ANOMALY DETECTION SYSTEM WITH POWER BI FOR VISUALIZATION OF ANALYTICAL DATA OF THE UNIVERSITY

Abstract

This article addresses the pressing issue of enhancing management efficiency in higher education institutions through the implementation of modern data analysis methods. It presents a practical architectural solution that integrates machine learning algorithms for anomaly detection with the powerful visualization capabilities of the Power BI platform. The study's objective is to automate the identification of deviations in educational, administrative, and operational processes within a university, thereby increasing the transparency and responsiveness of managerial decision-making. The research analyzed key metrics such as student performance, attendance, faculty workload, and classroom utilization. For anomaly detection, the study employed the Isolation Forest and DBSCAN algorithms, which are well-suited for handling noisy and sparse data. The results of a pilot implementation at a real faculty confirmed the system's high effectiveness: it identified instances of systematic faculty overload, sharp declines in attendance, inefficient use of premises, and delays in administrative procedures. Visualizing these anomalies in interactive Power BI dashboards made the analytics accessible to managers without deep technical expertise. The system's modular architecture allows for its scaling to various educational institutions and expanding its functionality by connecting new data sources and implementing predictive analytics.