IMPLEMENTATION OF DEEP LEARNING METHOD FOR INFLATION PROJECTIONS IN SOUTH SUMATRA PROVINCE

• Main Author: Wresti Buana Putri, Fany
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/63862
• Institution: Institut Teknologi Bandung

Low and stable inflation is a prerequisite for sustainable economic growth. In formulating and implementing monetary policy to maintain macroeconomic stability, inflation forecasts play a key role considering the lag of effect of monetary policy on inflation. Traditional statistical methods such as Autoregressive Integrated Moving Average (ARIMA) are popular methods used in forecasting and analyzing time-series data. However, the existence of limitations such as the assumption of models that are stationary and linear becomes an obstacle to this method to be able to predict economic problems that are generally nonlinear. Artificial neural network (ANN) models were introduced as a new approach to forecasting. In this study, a neural network model will be built to predict the inflation rate of South Sumatra Province. The result shows that the ANN model with a single layer was able to provide better South Sumatra’s inflation prediction results compared to the ARIMA model.