Time Series Analysis of Road Traffic Accidents in Ghana
Frank B. K. Twenefour *
Department of Mathematics, Statistics and Actuarial Science, Takoradi Technical University, P.O.Box 256, Takoradi, Ghana.
Emmanuel Ayitey
Department of Mathematics, Adiembra Senior High School, P.O.Box 16, Sekondi, Western Region, Ghana.
Justice Kangah
Department of Mathematics, Shama Senior High School, P.O.Box 30, Shama, Western Region, Ghana.
Lewis Brew
Department of Mathematics, University of Mines and Technology, P.O.Box 237, Tarkwa, Ghana.
*Author to whom correspondence should be addressed.
Abstract
This study uses Time Series models to predict the annual traffic accidents in Ghana. The traffic accidents data spanning from January 1990 to December 2019 was used. The Box-Jenkins model building strategy was used. The Augmented Dickey Fuller (ADF) test showed that the accident data was stationary. Three ARMA models were suggested based on the ACF and PACF plots of the differenced series, these were ARMA (0,0), ARIMA (1,0), and ARMA (2,0). The model with the smallest corrected Akaike Information Criteria (AICs) and Bayesian Information Criteria (BIC) was chosen as the best model. The Ljung-Box statistics among others were used in assessing the quality of the model. ARMA (1,0) was the best model for the Ghana annual Traffic Accident data. The results showed that, from January to July, it would be difficult to make accurate estimates of the number of road incidents for the years leading up to 2020. This was due to the fact that the white noise process values were statistically independent at various times.
Keywords: ARMA, Box-Jenkins model, ACF, ADF, AIC, BIC, PACF, Ghana annual traffic accident data, time series model.