Predicting Consumer Price Index amidst uncertainty: Gaussian Random Fuzzy Number-based Evidential Neural Network for West African economies with COVID-19 and Russia–Ukraine war dynamics

The accuracy of predicting Consumer Price Index (CPI) in West African economies is a complex issue influenced by various factors, including COVID-19 and ongoing Russia–Ukraine war. The current study examined the effectiveness of three different models, including Autoregressive Integrated Moving Average (ARIMA), Extreme Learning Machine (ELM), and Evidential Neural Network with Gaussian Random Fuzzy Numbers (EVNN-GRFN), in predicting the CPI for four West African nations: Ghana, Guinea Bissau, Nigeria, and Togo. The study use dummy variables to capture the periods of COVID-19 and Russia–Ukraine war. The EVNN-GRFN model demonstrated superior performance compared to the ARIMA model in terms of prediction accuracy. However, the incorporation of information about COVID-19 and the Russo-Ukrainian war had a varying impact on the performance of the EVNN-GRFN model, depending on the country. EVNN-GRFN significantly enhanced prediction accuracy by 12.93%, 7.14%, and 16.96% for Guinea Bissau, Nigeria, and Togo, respectively, compared to ARIMA. While incorporating information about COVID-19 and the Russo-Ukrainian war worsened the predictive accuracy for Guinea Bissau and Nigeria for the EVNN-GRFN model, it improved the accuracy by 7.70% for Ghana and 1.12% for Togo. This information only improved the accuracy of ELM for Guinea Bissau by 6.51%, but it worsened accuracy for Ghana, Nigeria, and Togo. Overall, the findings suggest that the EVNN-GRFN model is a promising tool for predicting CPI in West African economies and can be used to inform policy decisions regarding economic integration, monetary policy, and investment.