Abstract
In recent decades, cyber-physical systems (CPS) have become an essential part of modern industry and daily life. These systems integrate physical processes with computer and network components, allowing them to interact with their environment and manage their components autonomously. One of the most significant aspects of CPS efficiency is managing energy consumption, which significantly affects their reliability, efficiency, and economic performance. CPS devices generate vast amounts of diverse data, which is crucial to accurately model. Researchers use predictive analysis to develop models that forecast trends and simulate real-world conditions, enabling them to make better-informed decisions. This article presents a comparative analysis of different predictive models for CPS data analytics, focusing on energy consumption in smart buildings. Short-term models include gradient-boosted regressor (XGBoost), random forest (RF) and long short-term memory (LSTM). The comparative results have been studied in terms of prediction errors to determine accuracy.