PROFCAD: An Algorithm to Detect Anomalies in Cloud Applications for KPI Monitoring Systems
Rameshwar Garg, Chandana Kiran Ambekar, Kaustuv Saha, Girish Rao Salanke N.S. · 2021
Cloud based applications have a lot of advantages over traditional applications in terms of usability and cost, and have become the norm in today’s technological world. With the emergence of such services, it has become critical to make sure that they guarantee high availability. Analysing these applications and their performance indicators is one way to make sure that they are running smoothly. Machine learning techniques such as anomaly detection can be used to make sure that the Key Performance Indicators are behaving normally. In this paper, we propose a novel algorithm based on supervised learning to detect and identify anomalies in KPI data. PROphet Forest Combination for Anomaly Detection is a three stage model, based on forecasting, feature engineering and classification. We evaluate the performance of the model with two time series datasets which capture real traffic communications. Our model has been able to detect anomalies accurately and has performed well in comparison with the other state of the art anomaly detection algorithms.