Forecasting of Currency Circulation in Indonesia Using Hybrid Extreme Learning Machine

Abstract

Forecasting currency circulation, including inflow and outflow, is one of Bank Indonesia's strategies to maintain the Rupiah value's stability. The characteristic of inflow and outflow data is that they have seasonal variations. This study proposes a hybrid model by combining decomposition techniques and Extreme Learning Machine to overcome data that has seasonal variations. The forecasting results of the proposed model are compared with the original Extreme Learning Machine. The comparison results show that the forecasting results with the hybrid model have the smallest errors. Thus, the hybrid model can predict data with seasonal variations better than the original Extreme Learning Machine.

Forecasting the Indonesian Coffee Production and Consumption Using the Modified Golden Section Search to Estimate the Smoothing Parameters

“…A smaller MAPE value indicates a better forecasting process [41,42]. The MAPE is calculated with the following equation [43] M…”

Forecasting the Indonesian Coffee Production and Consumption Using the Modified Golden Section Search to Estimate the Smoothing Parameters

“…A smaller MAPE value indicates a better forecasting process [41,42]. The MAPE is calculated with the following equation [43] M…”