Processing IoT Sensor Fire Dataset Using Machine Learning Techniques

Abstract

Automatic fire detection now-a-days is important in Smart Home Management. Here in this study, popular machine learning algorithms viz. Support Vector Machine (SVM), Naïve Bayes, Decision Tree, Random Forest and K-Nearest Neighbor algorithm are employed on the dataset obtained from the Integrated Sensor Framework (ISF), designed with IoT sensors kit. The various classifiers shows various accuracy levels of fire sensor dataset of which SVM with polynomial kernel, SVM_RBF and Random forest provide high accuracy rate compared to other classifiers. The performance results show that the polynomial and RBF kernels of SVM and Random forest algorithms can predict cases of fire outbreak with an accuracy of 84% and a minimal error rate of 0.2%. The paper also highlights that the final outcome of the proposed approach will minimize error rate, thus making the system more reliable and accurate.