On context awareness and analysis of various classification algorithms

Abstract

Internet of Things (IoT) is currently connecting 9 billion devices and is expected to grow by three times in next 5 years, and hence will connect over 27 billion devices. IoT is touching every walk of human life such as health care, smart utilities, smart grid, smart homes, and smart spaces. To make things or object smart, IoT middleware makes use of appropriate intelligent mechanisms. Context-aware solutions are addressing the challenges of IoT middleware, hence becoming an important building block. We provide an analytical study of various algorithms for classification. We consider three algorithms and test the performances of each on small dataset as well as on larger dataset with 1969 instances. Performance evaluation is done using Mean Square Error and Absolute Mean Square Error.