"Assessment and Comparison of Ensemble Learning Techniques using Feature Reduction for Classification of IoT data"

Authors

Vijay M. Khadse (Assistant Professor)
College Of Engineering Pune (COEP), Pune, India.

Dr. Parikshit N. Mahalle (Senior Member IEEE, Professor & Head)
Smt. Kashibai Navale College of Engineering, Pune India. Post Doc Researcher, Center for Communication,
Media and Information
Technologies (CMI), Aalborg University, Copenhagen, Denmark.

Dr. Gitanjali R. Shinde (Assistant Professor)
Smt. Kashibai Navale College of Engineering, Pune India.

Abstract

In recent years, the growth of Internet of Things (IoT) as a prominent technology has been incredible. The number of network and sensor enabled devices in IoT domains is growing extremely, leading to the huge production of data. These data contain important information which can be used in various areas, such as science, industry, medical and even social life. To make IoT system smart, the only solution is entering the world of machine learning. Many machine learning algorithms are introduced and for handling such vast amount of IoT data. It is very difficult to find the best suited algorithm for problem in IoT domain. This study explores bagging, boosting and stacking ensemble learning models. This study combined three ensemble models and proposed a hybrid model. A set of features extracted from the raw IoT datasets using Principal component analysis (PCA), Linear discriminant analysis (LDA) and Isomap are used for classification. A performance comparison of the classifiers is provided in terms of their accuracy, area under the curve (AUC) and F1 score. The experimental results of this comparative study show that Hybrid with PCA and Stacking with PCA have better overall performance than other ensemble models for binary and multivariate datasets respectively,