Track: Big Data and Analytics

Abstract

The objective of this research is to analyse the deaths of people during Songkran holidays and to develop a model for the classification of deaths caused by road accidents. Data used in this research studies including Injuries and the loss of life in the accident between 2008 and 2014, a total of 2,875 people from the database of the Digital Government Development Office. The statistics used in the hypothesis testing are the Chi-square test statistic, Independent variables are behavior factors: drinking, not wearing a helmet, physical environmental factors such as the time when the road accident occurs, the type of road that caused the accident and the dependent variable was the death and injured person. The hypothesis testing at the significance level of 0.05 showed that all variables are associated with death during Songkran holidays. In addition, data mining techniques are applied to this research. Decision Tree, Bayesian Learning, Logistic Regression and Neural Network are applied to identify deaths described by a set of attributes and compare the accuracy of data classification with various data mining techniques. As the result, it was found that logistic regression can be correctly classified higher than other classification techniques with a precision of 72.20%.