ARIMA METHOD IN PREDICTING THE RUPIAH EXCHANGE RATE: THE EFFECT OF DAILY AND MONTHLY DATA FREQUENCY ON THE ACCURACY OF SHORT-TERM PREDICTIONS

Abstract

The Rupiah exchange rate against the US Dollar (USD) serves as a critical economic indicator influenced by internal and external factors. In time series analysis, the ARIMA (Autoregressive Integrated Moving Average) model is often used to predict the exchange rate. This study examines how data frequency daily versus monthly affects the accuracy of ARIMA predictions for the Rupiah exchange rate. Using historical exchange rate data from Bank Indonesia, this study found that daily data produces more accurate predictions than monthly data based on a comparison of the forecast value with the actual exchange rate on that day. This finding suggests that high volatility in daily data reduces ARIMA's ability to capture short-term patterns, while monthly data provides a more stable pattern for medium-term predictions. This study provides insights for economists, researchers, and practitioners in determining the optimal data frequency for currency exchange rate predictions.