A research team from Universitas Padjadjaran has developed an advanced machine learning model to forecast stock prices of PT Indika Energy Tbk (INDY), a major player in Indonesia’s energy sector. The study was recently published in Mathematics (Louisa et. al, 2025).
The hybrid model, called CNN-BiGRU-Attention (CNN-BiGRU-AM), combines Convolutional Neural Networks (CNN), Bidirectional Gated Recurrent Units (BiGRU), and an Attention Mechanism (AM) to handle the nonlinear, volatile, and noisy nature of stock market data. By analyzing historical data from August 2019 to February 2025, the model successfully predicted stock price movements with high accuracy, achieving a Mean Absolute Percentage Error (MAPE) below 3%.
The CNN component extracts local patterns and reduces noise, BiGRU captures temporal dependencies in both directions, and the Attention Mechanism highlights the most relevant historical information. This allows companies, investors, and policymakers to make more informed strategic decisions, even amid global commodity price fluctuations and geopolitical events.
Beyond financial applications, this research contributes to the Sustainable Development Goals (SDGs) by supporting:
- SDG 8 (Decent Work and Economic Growth): enhancing informed economic and investment decisions.
- SDG 9 (Industry, Innovation, and Infrastructure): applying advanced AI technology to financial modeling.
- SDG 17 (Partnerships for the Goals): encouraging collaboration among academia, industry, and stakeholders.
This innovative approach demonstrates how cutting-edge AI can improve financial forecasting in complex and dynamic markets while promoting sustainable economic development in Indonesia.
Source: https://www.scopus.com/record/display.url?eid=2-s2.0-105010340720&origin=resultslist
23/Stat/2025
