MODELLING ROAD ACCIDENTS IN SELANGOR: A COMPARATIVE ANALYSIS BY USING SEASONAL AUTOREGRESSIVE INTEGRATED MOVING AVERAGE (SARIMA) AND ARTIFICIAL NEURAL NETWORK (ANN)

• Published in: Journal of Sustainability Science and Management
• Main Author: Azan W.N.A.W.M.; Borhan N.; Zahari S.M.
• Format: Article
• Language: English
• Published: Universiti Malaysia Terengganu 2023
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85160649358&doi=10.46754%2fjssm.2023.05.007&partnerID=40&md5=7198e3da63f411623e5f8fcce9edd5cf

Road accidents have become one of the major problems globally. It was reported that the total number of road accidents in Malaysia has increased by 0.92 per cent within the last ten years. Modelling road accident occurrences is critical for policymakers to understand the trend and pattern of road accidents to provide an appropriate countermeasure. This study attempts to forecast the road accident occurrences on federal and state roads in Selangor, Malaysia. This study utilised monthly road accident data from January 2011 to December 2021. The traditional univariate Seasonal Autoregressive Integrated Moving Average (SARIMA) and Artificial Neural Network (ANN) models were employed and the performance of each model was assessed. The findings found that the ANN model outperformed the SARIMA model in training and validation. Furthermore, the ANN model has the lowest RMSE, MAE, MAPE and MASE values for both sets. This study demonstrates the potential of machine learning in forecasting and predicting road accident occurrences, giving more flexibility and assumption-free methodology. © Penerbit UMT