A Hybrid Singular Spectrum Analysis and Neural Networks for Forecasting Inflow and Outflow Currency of Bank Indonesia
This study proposes hybrid methods by combining Singular Spectrum Analysis and Neural Network (SSA-NN) to forecast the currency circulation in the community, i.e. inflow and outflow. The SSA technique is applied to decompose and reconstruct the time series factors which including trend, cyclic, and seasonal into several additive components, i.e. trend, oscillation and noise. This method will be combined with Neural Network as nonlinear forecasting method due to inflow and outflow data have non-linear pattern. This study also focuses on the effect of Eid ul-Fitr as calendar variation factor which allegedly affect inflow and outflow. Thus, the proposed hybrid SSA-NN is evaluated for forecasting time series that consist of trend, seasonal, and calendar variation patterns, by using two schemes of forecasting process, i.e. aggregate and individual forecasting. Two types of data are used in this study, i.e. simulation and real data about the monthly inflow and outflow of 12 currency denominations. The forecast accuracy of the proposed method is compared to ARIMAX model. The results of the simulation study showed that the hybrid SSA-NN with aggregate forecasting yielded more accurate forecast than individual forecasting. Moreover, the results at real data showed that the hybrid SSA-NN yielded as good as ARIMAX model for forecasting of 12 inflow and outflow denominations. It indicated that the hybrid SSA-NN could not successfully handle calendar variation pattern in all series. In general, these results in line with M3 competition conclusion, i.e. more complex methods do not always yield better forecast than the simpler one.
KeywordsSingular spectrum analysis Neural network Hybrid method Inflow Outflow
This research was supported by DRPM-DIKTI under scheme of “Penelitian Berbasis Kompetensi”, project No. 851/PKS/ITS/2018. The authors thank to the General Director of DIKTI for funding and to anonymous referees for their useful suggestions.
- 1.Sigalingging, H., Setiawan, E., Sihaloho, H.D.: Money Circulation Policy in Indonesia. Bank Indonesia, Jakarta (2004)Google Scholar
- 2.Apriliadara, M., Suhartono, A., Prastyo, D.D.: VARI-X model for currency inflow and outflow with Eid Fitr effect in Indonesia. In: AIP Conference Proceedings, vol. 1746, p. 020041 (2016)Google Scholar
- 6.Broomhead, D.S., King, G.P.: On the qualitative analysis of experimental dynamical systems. In: Sarkar S (ed.) Nonlinear Phenomena and Chaos, pp. 113–144. Adam Hilger, Bristol (1986)Google Scholar
- 7.Broomhead, D.S., Jones, R., King, G.P., Pike, E.R.: Singular Spectrum Analysis with Application to Dynamic Systems, pp. 15–27. IOP Publishing, Bristol (1987)Google Scholar
- 10.Zhigljavsky, A., Hassani, H., Heravi, S.: Forecasting European Industrial Production with Multivariate Singular Spectrum Analysis. Springer (2009)Google Scholar
- 12.Li, H., Cui, L., Guo, S.: A hybrid short term power load forecasting model based on the singular spectrum analysis and autoregressive model. Adv. Electr. Eng. Artic. ID 424781, 1–7 (2014)Google Scholar
- 13.Lopes, R., Costa, F.F., Lima, A.C.: Singular spectrum analysis and neural network to forecast demand in industry. In: Brazil: The 2nd World Congress on Mechanical, Chemical, and Material Engineering (2016)Google Scholar
- 14.Sun, M., Li, X., Kim, G.: Precipitation analysis and forecasting using singular spectrum analysis with artificial neural networks. Clust. Comput., 1–8 (2018, in press)Google Scholar
- 19.Lee, M.H., Suhartono, A., Hamzah, N.A.: Calendar variation model based on ARIMAX for forecasting sales data with Ramadhan effect. In: Regional Conference on Statistical Sciences, pp. 349–361 (2010)Google Scholar
- 21.Suhartono: New procedures for model selection in feedforward neural networks. Jurnal Ilmu Dasar 9, 104–113 (2008)Google Scholar
- 22.Crone, S.F., Kourentzes, N.: Input-variable specification for neural networks - an analysis of forecasting low and high time series frequency. In: International Joint Conference on Neural Networks, pp. 14–19 (2009)Google Scholar