Time Series Analysis Of Motor Vehicle Accidents In Zimbabwe (1993-2023). A Comparative Study Of Autoregressive Integrated Moving Average And Artificial Neural Network Models
- Author
- Tanyaradzwa Zvomuya
- Title
- Time Series Analysis Of Motor Vehicle Accidents In Zimbabwe (1993-2023). A Comparative Study Of Autoregressive Integrated Moving Average And Artificial Neural Network Models
- Abstract
- This study is a comprehensive time series analysis of yearly motor vehicle accidents in Zimbabwe from 1993 to 2023, comparing the performance of Multilayer Perceptron model (MLP) and the Autoregressive Integrated Moving Average (ARIMA). The aim was to identify the most accurate and reliable modelling approach for forecasting number of motor vehicle accidents in Zimbabwe. A quantitative research design for 31 data points of yearly motor vehicle accidents was employed in the study and the R-Software was used to perform data analysis. ARIMA model was developed using the Box-Jenkins model building strategy. The Augmented Dickey Fuller test revealed that the accident data was non-stationary. After first order differencing, the data became stationary. The model with the smallest corrected Akaike Information Criteria (AICc) and Bayesian Information Criteria (BIC) was chosen as the best model which is the ARIMA (1,0,0) model. The best-performing MLP model among the three that were created was 1-(10,5)-1. The performance evaluation metrics were used to compare the models against observed data from 2017 to 2023. Mean Absolute Error and Root Mean Square Error, were used as performance evaluation metrics. This study's conclusions indicate that MLP out performed ARIMA model and it was used for forecasting number of future accidents for the next 5 years. Future projections indicate a downward trend in the number of motor accidents. Even if the trend in future accidents was decreasing there is still a need to exercise more caution so as to reduce the occurrence of accidents related to road traffic crashes.
- Date
- June 2024
- Publisher
- BUSE
- Keywords
- Integrated Moving Average
- Artificial Neural
- Network Models
- Supervisor
- DR T. W. MAPUWEI
