Analysis of Driver’s Attention through the Internet of Things (IOTs) for Preventing Road Accident of Natural Gas Vehicles

Abstract

The objectives of this study were the following: (1) to investigate the correlations between data collected through Internet of Thing (IOT) and unintentional behavior of drivers (2) to create the models based on machine learning techniques to classify unintentional behavior of drivers who drive the natural gas vehicle and (3) to compare the forecasting accuracy of the learning model. Data studied were collected from the system of the natural gas transportation business in Thailand. There were 10,693 records starting from January 1, 2019 to December 31, 2019, for a period of 12 months. Moreover, KNIME Analytics Platform was used to create the model. The research findings were as follows: (1) duration time when the driver is not looking straight, driving speeds, distance coverage of the driver faces that is not looking straight detecting by a camera and the latitude and longitude coordinates have a relationship with unintentional behavior of the driver; and (2) Neural Network with two hidden layer and 5 neurons in the hidden layer performs the highest accuracy (873%), followed by Support Vector Machine with S3.9%, of accuracy. It can be said that Neural Network can be used to create an efficient predictive model.