MODEL PERAMALAN NILAI TUKAR RUPIAH TERHADAP DOLLAR SINGAPURA MENGGUNAKAN METODE HYBRID ARIMA-ANN

Abstract

This research aims to predict the Rupiah exchange rate against the Singapore Dollar using the hybrid ARIMA-ANN method. The hybrid model is used to increase prediction accuracy by utilizing the ARIMA model to capture linear patterns and the ANN model to capture non-linear patterns. The data used in this research is data on the Rupiah exchange rate against the Singapore Dollar. The ARIMA model used for hybrid modeling is ARIMA (1,1,1) because it has an AIC value of 2144.93 which is smaller than other ARIMA models. The residuals from the ARIMA model (1,1,1) are used for ANN modeling. ANN modeling uses 3 inputs, 1-10 hidden layers, and 1 output layer. Based on the analysis results, the ARIMA (1,1,1) - ANN (3,10,1) hybrid model has an RMSE value of 52.092 which is smaller than other ARIMA-ANN hybrid models. Therefore, the hybrid ARIMA (1,1,1) - ANN (3,10,1) model is more effective in predicting the Rupiah exchange rate against the Singapore Dollar.

Keywords

• arima, ann, time series, exchange rate.
• deret waktu, nilai tukar