TY - JOUR
T1 - Forecasting Inflow and Outflow of Currency in Central Java using ARIMAX, RBFN and Hybrid ARIMAX-RBFN
AU - Maghfiroh, Z. F.
AU - Suhartono,
AU - Prabowo, H.
AU - Salehah, N. A.
AU - Prastyo, D. D.
AU - Setiawan,
N1 - Publisher Copyright:
© Published under licence by IOP Publishing Ltd.
PY - 2021/4/19
Y1 - 2021/4/19
N2 - This research aims to forecast the inflow and outflow currency in Central Java. Inflow and outflow data contained both non-linear and linear patterns with calendar variation effects. Calendar variation model based on ARIMAX as a linear model, Radial Basis Function Network (RBFN) as a non-linear model, and hybrid ARIMAX-RBFN as a combination linear and non-linear model are used to forecast inflow and outflow of currency in Central Java. The data used in this research consists of inflow and outflow of currency in Central Java from January 2010 until June 2019. The denomination used is 32 denominations of inflow and 32 denominations of outflow currency. RMSE and sMAPE values from the out-of-sample data are used to select the best model. The results show that hybrid ARIMAX-RBFN is the best model of 19 denominations of inflow currency and 22 denominations of outflow. In general, the hybrid model tends to provide a more accurate forecast than the individual forecasting model used in this research.
AB - This research aims to forecast the inflow and outflow currency in Central Java. Inflow and outflow data contained both non-linear and linear patterns with calendar variation effects. Calendar variation model based on ARIMAX as a linear model, Radial Basis Function Network (RBFN) as a non-linear model, and hybrid ARIMAX-RBFN as a combination linear and non-linear model are used to forecast inflow and outflow of currency in Central Java. The data used in this research consists of inflow and outflow of currency in Central Java from January 2010 until June 2019. The denomination used is 32 denominations of inflow and 32 denominations of outflow currency. RMSE and sMAPE values from the out-of-sample data are used to select the best model. The results show that hybrid ARIMAX-RBFN is the best model of 19 denominations of inflow currency and 22 denominations of outflow. In general, the hybrid model tends to provide a more accurate forecast than the individual forecasting model used in this research.
UR - http://www.scopus.com/inward/record.url?scp=85104789775&partnerID=8YFLogxK
U2 - 10.1088/1742-6596/1863/1/012066
DO - 10.1088/1742-6596/1863/1/012066
M3 - Conference article
AN - SCOPUS:85104789775
SN - 1742-6588
VL - 1863
JO - Journal of Physics: Conference Series
JF - Journal of Physics: Conference Series
IS - 1
M1 - 012066
T2 - International Conference on Mathematics, Statistics and Data Science 2020, ICMSDS 2020
Y2 - 11 November 2020 through 12 November 2020
ER -