CONTACT US: Contact info
|
Citation |
Mphale O, Lakshmi Narasimhan V, Agrawal S, Gupta KK, Chan JH, Agrawal J, Gupta M. Algoritm. Intel. Syst. 2022; 309-328. |
|
Copyright |
(Copyright © 2022, Holtzbrinck Springer Nature Publishing Group) |
|
DOI |
|
PMID |
unavailable |
|
Abstract |
Road traffic accidents (RTAs) are the major cause of deaths worldwide, which imposes a substantial economic burden on the society. RTAs are triggered by several factors that can manifest individually or in association resulting in discernible patterns. This paper employs data mining approach, namely association rule mining algorithm, to discover latent relationships between various factors triggering RTAs in Botswana. The findings show that most traffic accidents in Botswana occur in Kweneng, Central and South East districts. Of these, 69% are minor, 22% serious and 9% fatal accidents. Furthermore, these accidents occur often on Saturdays and Fridays, with about 20% and 15% casualties involved, respectively. The drivers are mostly young males who are recurrently under the influence of alcohol or drugs. Furthermore, 99% of RTAs drivers hold valid driver's licences and 61% often collide with pedestrians crossing the road in daylight. The primary critical factors accounting to RTAs are alcohol/drugs, driver age, driver licence validity, weather, road curvature and road surface type, which relate to the driver attribute class and the accident attribute class congruently. The findings reveal that there are several strong relationships among the factors, such as the driver licence validity, driver alcohol/drugs, victim, weather, accident severity and road condition. These factors, show positive correlations in their co-occurrences in unique traffic accidents events in Botswana. Language: en |
|
Keywords |
Association rule mining; Data mining; FP-Growth algorithm; Knowledge discovery; Traffic accident data; Visualisation |