With the advent of 5G technologies, IoT infrastructures are the key enablers in providing smart solutions like smart homes, smart industrial infrastructure, and smart cities. With the increasing availability of Internet of Things (IoT) devices as an interconnected network, smart homes have emerged as promising solutions for enhancing energy efficiency by intelligently regulating energy consumption. Traditional home energy management systems rely on predefined schedules or manual user interventions to regulate energy usage. This limitation restrains the realization of substantial energy savings from a domestic perspective. To overcome these challenges, our study proposes an algorithm IPA-DS. IPA- DS estimates the current energy usage statistics in real-time home environment. The estimated statistics are then exploited as features to train the machine learning algorithms. The trained algorithms are then tested and validated using error functions and processing overheads.