A Short-Term Prediction Model for the Number of Registered Motor Vehicles Using Facebook Prophet Forecasting Approach

Accurate prediction of the total annual number of motor vehicles to be registered in a country is very important to the government because it will help Driver and Vehicle Licensing Authority (DVLA) policymakers to put good road safety measures in place to make their work successful. This study assessed four standard empirical time series methods for the prediction of the total number of motor vehicles to be registered. The evaluated methods were Facebook Prophet, grey, SARIMA, and ARIMA. In developing the time series model, 70% of the data set obtained from DVLA of Ghana was used and the remaining 30% was used to validate the model’s forecasting adequacy. The robustness of the Facebook Prophet, grey, SARIMA, and ARIMA was assessed using Normalised Root Mean Square Error (NRMSE), Correlation Coefficient (R), Variance Accounted For (VAF), and Performance Index (PI). The analytical results revealed that the Facebook Prophet time series model had the best results with the lowest NRMSE (0.2176) and highest R (0.8229), VAF (66.6475%), and PI (1.1260) values. The study concluded that the Facebook Prophet forecasting approach could be useful to vehicle licensing authority managers and policymakers to give accurate and timely information on future projections of the number of motor vehicles to be registered.