Internet of Things Flood Prediction Architecture using Naive Bayes, Non- Linear Support Vector Machines (SVM) and Multi-Layer Perceptron (MLP) Classifiers

  • Kovvuri N. Bhargavi, Dr. G. Jaya Suma

Abstract

Disaster Management deals with awareness, response and recovery in emergency situations. Natural disasters like floods are difficult to predict as many uncertain attributes are needed to be analyzed. Prediction of Floods is however virtually impossible. Internets of Things (IoT) devices provide a way to monitor the affected areas regularly and measure its attributes for the prediction using sensors to sense real time. Many existing methods were applied in the flood prediction that has low scalability and efficiency in the forecasting. In this research, an IoT architecture is developed for monitoring the particular area and calculate its attributes for the prediction using machine learning techniques. The non-linear SVM analyzes the attribute and predicts the disaster effectively due to the Radial Basis Function (RBF) method. The uncertain attributes are analyzed effectively by the Naive Bayes and non-linear SVM for flood prediction. The prediction analysis is carried out using Naïve Bayes, Non-Linear SVM, and MLP. The experimental result shows that the Naive Bayes model with the IoT structure method has a higher performance in the flood forecasting than other machine learning models.

Published
2020-03-30
How to Cite
Kovvuri N. Bhargavi, Dr. G. Jaya Suma. (2020). Internet of Things Flood Prediction Architecture using Naive Bayes, Non- Linear Support Vector Machines (SVM) and Multi-Layer Perceptron (MLP) Classifiers. International Journal of Advanced Science and Technology, 29(3), 12205 - 12216. Retrieved from https://sersc.org/journals/index.php/IJAST/article/view/30311
Section
Articles