A Comparison of Forecast Behavior of Four Time Series Models on Nigeria’s External Reserves
Article Sidebar
Main Article Content
Abstract
This article compared decomposition, Holt-Winters, ARIMA and Artificial neural networks (ANN) models for forecast accuracy and similarity on Nigeria’s external reserves. Fitted adequate ARIMA model is ARIMA (1, 1, 1) while ANN (12, 7, 1) possessing minimum mean absolute error (MAE) of six ANN models fitted was chosen for forecasting. Holt-Winters model utilized parameter values 0.9, 0.3 and 0.3 for level, trend and season respectively. The parameters were selected such that MAE was minimal. Based on forecast mean square error of twelve out-of-sample forecasts, ARIMA performed best, followed by Holt-Winters and then, decomposition. ANN performed worst. Fitted ARIMA suggests an increasing trend for the series. With increasing lead time, Holt-Winters forecasts substantially tend to move in an opposite direction to each of decomposition and ARIMA forecasts. The article observes that for Nigeria’s external reserves to attain high levels of the past there must be improved governance, security and infrastructure. The need to extend study to other economic series was recommended.
Downloads
Article Details
References
Adebiyi, A.A., Adewumi, A. O & Ayo, C.K. (2014). Comparison of ARIMA and Artificial Neural Networks models for stock price prediction. Journal of Applied Mathematics, 2014, Article Id 6114342. https://doi.org/10.1155/2014/614342
Amaefula, C.G. (2021). A SARIMA and adjusted SARIMA models in a seasonal nonstationary time series; Evidence of Enugu monthly rainfall. European Journal of Mathematics & Statistics, 2(1), 13-18. https//doi.org/10.24018/ejmath.2021.2.1.15
Bolarinwa, B.T. & Bolarinwa, I.A. (2021a). Single Versus Combined Forecasting: The Case of Wind Run Data. European Scientific Journal, ESJ, 17(21), 189. https://doi.org/10.19044/esj.2021.v17n21p189
Bolarinwa, B.T. & Bolarinwa, I.A. (2021b). An investigation of the performance of three forecasts combining schemes on solar radiation data. International Journal of Mathematics and Statistics Studies, 9(1), 41-47.
Bolarinwa, I.A. & Bolarinwa, B.T. (2021c). Understanding features of time series through exponential smoothing. International Journal of Advanced Academic Research, 7(8), 15-19. DOI: 10.46654
Bolarinwa, I.A. & Bolarinwa, B.T. (2021d). A suggested approach to artificial neural networks modeling of time series. European Journal of Statistics and Probability, 9(1), 11-16.
Central Bank of Nigeria (2020). Statistical Bulletin.: Central Bank of Nigeria.
Etuk, E.H., Aboko, I.S., Victor-Edema, U. & Dimkpa, M.Y. (2014). An additive seasonal Box-Jenkins model for Nigerian monthly savings deposit rates. Issues Bus. Manag. Econ. 2(3):054-059.
Kandananond, K. (2012). A comparison of various forecasting methods for autocorrelated time series. International Journal of Engineering Business Management. DOI: 10.5772/51088.
Merh, N., Saxena, V.P. & Pardasani, K.R. (2010). A comparison between hybrid approaches of ANN and ARIMA for Indian stock trend forecasting. Journal of Business Intelligence, 3(2), 23-43.
Nyaboke, C, Orwa, G.O. & Joseph, M. (2017). Time series modeling with application to Kenyan’s inflation data comparison of ARIMA and ARCH models. International Journal of Statistics and Applied Mathematics, 2(6), 16-22.
Otu, A., Osiju, G., Jude, O. & Ifeyinwa, M.H. (2014). Application of Sarima in modeling and forecasting Nigeria’s inflation rates. American Journal of Applied Mathematics and Statistics, 2(1), 16-28. Doi.10.12691/ajams-2-1-4