TY - GEN
T1 - Forecasting Road Deaths in Malaysia Using Support Vector Machine
AU - Radzuan, Nurul Qastalani
AU - Hassan, Mohd Hasnun Arif
AU - Majeed, Anwar P.P.Abdul
AU - Kassim, Khairil Anwar Abu
AU - Musa, Rabiu Muazu
AU - Razman, Mohd Azraai Mohd
AU - Othman, Nur Aqilah
N1 - Funding Information:
Acknowledgements The authors would like to acknowledge ASEAN NCAP, FIA Foundation, Global NCAP, OEMs and the Society of Automotive Engineers Malaysia (SAE Malaysia) for funding this study under the ASEAN NCAP Holistic Collaborative Research (ANCHOR II) grant (UIC191504). Also, the authors are thankful to the Universiti Malaysia Pahang for providing the facilities to conduct the study.
Funding Information:
The authors would like to acknowledge ASEAN NCAP, FIA Foundation, Global NCAP, OEMs and the Society of Automotive Engineers Malaysia (SAE Malaysia) for funding this study under the ASEAN NCAP Holistic Collaborative Research (ANCHOR II) grant (UIC191504). Also, the authors are thankful to the Universiti Malaysia Pahang for providing the facilities to conduct the study.
Publisher Copyright:
© 2020, Springer Nature Singapore Pte Ltd.
PY - 2020
Y1 - 2020
N2 - An average of 6,350 people died every year in Malaysia due to road traffic accidents. A published data of Malaysian road deaths in 20 years since 1997 reveals that the number of fatalities has not really declined with a difference of less than 10% from one year to the next. Forecasting the number of fatalities is beneficial in planning a countermeasure to bring down the death toll. A predictive model of Malaysian road death has been developed using a time-series model known as autoregressive integrated moving average (ARIMA). The model was used in the previous Road Safety Plan of Malaysia to set a target death toll to be reduced in 2020, albeit being inaccurate. This study proposes a new approach in forecasting the road deaths, by means of a machine learning algorithm known as Support Vector Machine. The length of various types of road, number of registered vehicles and population were among the eight features used to develop the model. Comparison between the actual road deaths and the prediction demonstrates a good agreement, with a mean absolute percentage error of 2% and an R-squared value of 85%. The Linear kernel-based Support Vector Machine was found to be able to predict the road deaths in Malaysia with reasonable accuracy. The developed model could be used by relevant stakeholders in devising appropriate policies and regulations to reduce road fatalities in Malaysia.
AB - An average of 6,350 people died every year in Malaysia due to road traffic accidents. A published data of Malaysian road deaths in 20 years since 1997 reveals that the number of fatalities has not really declined with a difference of less than 10% from one year to the next. Forecasting the number of fatalities is beneficial in planning a countermeasure to bring down the death toll. A predictive model of Malaysian road death has been developed using a time-series model known as autoregressive integrated moving average (ARIMA). The model was used in the previous Road Safety Plan of Malaysia to set a target death toll to be reduced in 2020, albeit being inaccurate. This study proposes a new approach in forecasting the road deaths, by means of a machine learning algorithm known as Support Vector Machine. The length of various types of road, number of registered vehicles and population were among the eight features used to develop the model. Comparison between the actual road deaths and the prediction demonstrates a good agreement, with a mean absolute percentage error of 2% and an R-squared value of 85%. The Linear kernel-based Support Vector Machine was found to be able to predict the road deaths in Malaysia with reasonable accuracy. The developed model could be used by relevant stakeholders in devising appropriate policies and regulations to reduce road fatalities in Malaysia.
KW - Machine learning
KW - Prediction
KW - Road death
KW - Road traffic accident
KW - Support vector machine
UR - http://www.scopus.com/inward/record.url?scp=85083073913&partnerID=8YFLogxK
U2 - 10.1007/978-981-15-2317-5_22
DO - 10.1007/978-981-15-2317-5_22
M3 - Conference Proceeding
AN - SCOPUS:85083073913
SN - 9789811523168
T3 - Lecture Notes in Electrical Engineering
SP - 261
EP - 267
BT - InECCE 2019 - Proceedings of the 5th International Conference on Electrical, Control and Computer Engineering
A2 - Kasruddin Nasir, Ahmad Nor
A2 - Saari, Mohd Mawardi
A2 - Daud, Mohd Razali
A2 - Mohd Faudzi, Ahmad Afif
A2 - Ahmad, Mohd Ashraf
A2 - Najib, Muhammad Sharfi
A2 - Abdul Wahab, Yasmin
A2 - Othman, Nur Aqilah
A2 - Abd Ghani, Nor Maniha
A2 - Irawan, Addie
A2 - Khatun, Sabira
A2 - Raja Ismail, Raja Mohd Taufika
PB - Springer
T2 - 5th International Conference on Electrical, Control and Computer Engineering, InECCE 2019
Y2 - 29 July 2019 through 29 July 2019
ER -