A review of data analytic algorithms for outlier detection on the internet of things ecosystem

Abstract

In the last few years, outlier detection has drawn a lot of attention. New technologies, including the Internet of Things (IoT), are recognized as one of the most important sources of data streams, continuously producing enormous amounts of data from several applications. Reducing functional hazards and avoiding hidden problems that result in application downtime can be achieved by looking through this gathered data to identify suspicious events. This paper presents a review of existing algorithms deployed on Internet of things ecosystem that resolved the challenges of data outliers. It further highlights the problems solved, the results and the weaknesses of the existing algorithms. Also, presented a detailed discussion on various programming language and simulation tools adopted to implement and conduct experiment on the prevailing algorithms; as well as metrics used to evaluate their performances. It was discovered that metrics such as accuracy, precision, recall, specificity are mostly adopted as metrics for performance evaluation of the algorithms. Additionally, python programming language and Microsoft Studio IDE simulation tools were mostly used for the development and test-running of the existing algorithms.