An Evaluation of ARIMA and Persistence Models in IoT-Driven Smart Homes

Abstract

Commencing with the transformative fusion of Smart Home and Internet of Things (IoT) technologies, this study scrutinizes the efficacy of predictive modeling approaches, specifically the autoregressive integrated moving average (ARIMA) and persistence algorithms. The primary focus lies in their potential for forecasting and optimizing energy consumption dynamics within the intricate framework of smart homes. The investigation reveals a nuanced comparison between the proposed ARIMA and conventional Persistence models. Smart Home, emblematic of innovative living, integrates seamlessly with IoT, promising an intelligent and interconnected domestic ecosystem. To enhance energy efficiency, this study explores the ARIMA model's capabilities alongside the persistence algorithm. Notably, the proposed ARIMA model showcases exceptional prowess in forecasting, substantiated by a significantly lower compared to the Persistence model. The ARIMA model, with an Root Mean Square Error value of 0.03378, outshines the Persistence model with a higher Root Mean Square Error value of 0.158 when evaluated on the test dataset. This substantial reduction in emphasizes the superior performance of the ARIMA model, making it a compelling choice for time series forecasting tasks. Beyond quantitative metrics, the precision of the ARIMA model holds transformative potential, promising cost-effective energy consumption, proactive maintenance, and an elevated quality of life within smart homes. This research establishes a robust foundation for integrating advanced predictive modeling, particularly the ARIMA model, to enhance the efficiency, sustainability, and inhabitant satisfaction of smart homes in the era of IoT.