Intelligent Classification of IoT Traffic in Healthcare Using Machine Learning Techniques

This paper addresses the problem of improved classification of data available at IoT node using three well-established machine learning-based classifiers. The review of recent literature reveals that few work has been reported on classification and forecasting of IoT based data using machine learning techniques. On the other hand, in recent years, online classification and prediction of health care data is gaining importance. Keeping this motivation in view, in the present work, intelligent classification of Parkinson's disease using IoT based data has been carried out employing machine learning techniques. The machine learning-based classifiers used in this paper are Decision Tree, Random Forest, and Naive Bayes which are chosen based on their consistent and improved classification performance for other standard data sets. The IoT based node receives the data and offers the classification solution faster so that it helps in the decision-making. By using the two typical data sets, the simulation-based experiments are conducted. F1 score and execution time for both data sets for each classifier are obtained and compared. Further, the effect of the number of features on classification accuracy is studied from the simulation results. The results demonstrate that the ranking in terms of accuracy of classification is the Random Forest, Decision Tree and Naive Bayes. However, in terms of execution time, the ranking of the models are Naive Bayes, Decision Tree and Random Forest. Depending upon the requirement, the appropriate classifier needs to be selected to be used in IoT based industrial environments.