Application of Data Science in Analyzing Factors Related to Traffic Accidents in Thailand

Abstract

          The research of this were: 1) To study the development of a model for identifying traffic accident causation topics using the Latent Dirichlet Allocation (LDA) method, which helps in identifying key causal factors; and 2) To analyze the factors contributing to the risk of traffic-related fatalities in Thailand using a logistic regression algorithm, to provide guidance for government interventions in line with the Thailand Road Safety Master Plan (2022-2027) and reduce future traffic fatalities. The analysis is based on data from 2,943 news articles obtained from the online news platform of Amarin TV.
         The result showed the following. 1) The causal factors were categorized into three main topics: (1) individual factors, (2) vehicle-related factors, and (3) physical environment factors (roads/conditions); and 2) The analysis of factors contributing to traffic fatalities in Thailand identified four significant predictive variables: individual factors combined with vehicle-related factors, individual factors alone, vehicle-related factors alone, and a combination of individual, vehicle-related, and physical environment factors (roads/conditions). These factors have a significant influence on traffic fatalities in Thailand.