TIME SERIES MODELING USING ARIMA WITH INTERVENTIONS AND OUTLIERS FACTOR ALSO EXOGENOUS VARIABLES

• Main Author: Savero, Bryan
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/77021
• Institution: Institut Teknologi Bandung

Time series is a sequence of data observations measured over a specific period with equally spaced intervals. Time series analysis and forecasting play a crucial role in analyzing economic changes and facilitating effective decision-making. In this study, inflation and unemployment rates in Indonesia are used as the research objects. These data are affected by intervening factors such as the Covid-19 pandemic and the reduction of fuel subsidies by the 6th President of Indonesia, as well as the presence of outliers in the inflation data. Additionally, exogenous variables such as economic growth and interest rates are included in the ARIMAX modeling. The selection of ARIMA models with interventions and outliers, as well as ARIMAX, is based on the analysis and comparison of their performance. The research findings reveal that the ARIMA model with interventions provides better accuracy for unemployment data, with a MAPE of 5.91%. On the other hand, for inflation data, the ARIMAX model demonstrates the best performance with a MAPE of 7.97%. These results contribute significantly to understanding the factors influencing inflation and unemployment rates in Indonesia. By employing the ARIMA model with interventions and outliers, as well as ARIMAX, predictions can be made with improved accuracy. This has significant implications for economic decision-making and future policy planning.