An Efficient Machine Learning-Based Cluster Analysis Mechanism for IoT Data

Abstract

The prevailing developments in internet of things (IoT) and other sensor technologies such as cyber physical systems (CPS) and wireless sensor networks (WSNs), the huge amount of sensor data has been generating from various IoT devices and protocols. Making predictions and finding density patterns over such data is a challenging task. In order to find the density patterns and make analysis over real-time dynamic data, the machine learning (ML) based algorithms are widely used to deal with the IoT data. In this article, the authors proposed an efficient ML-based cluster analysis mechanism for finding density patterns in IoT dynamic data effectively. In this proposed mechanism, the k-means and GMM models are used for clustering data analysis. The proposed mechanism has been implemented on ThingSpeak Cloud platform for analysing the data efficiently on daily and weekly basis. Finally, the proposed mechanism acquired superior results than the existing benchmarked mechanisms over all the performance evaluation metrics used for analysis over IoT dynamic data.