Energy related activities recognition using smartphones

Abstract

Due to the immensely popularity of smartphones, more applications and functions are developed each day for the smartphone users. This creates an endless cycle as these applications are targeted at leisure, lifestyle, and work purposes, which users starts to grow reliance for. In order to accommodate this cycle, engineers are relied upon to develop and invent new hardware and software solutions. As a result, smartphones are now equipped with cutting-edge tools, high-speed processors, HD camera and ultra-reliable embedded sensors. This brings forward the notion of whether smartphones are now equipped to conduct recognition activities, and potentially enhance life as a result. This research aims to study the best method to achieve smart phone recognition of energy activities – namely movement activity and weather activity. In order to do so, 10 participants were invited to partake in the experiment to procure activity data on the smart phone. Then, it compares different machine learning algorithm, like KNN, SVM and Decision Tree based on their accuracy and implant this dataset into its Mobile Application. Lastly, with the assimilation of Government weather data, an AR system for the smart phone is finally created. The end product is a system that is accurate, robust and capable of making lifestyle-enhancing decisions for the user.