STOCK PRICE PREDICTION WITH TEXT MINING ALGORITHM AND LONG SHORT TERM MEMORY APPROACH: PT BUMI RESOURCES TBK STOCK CASE STUDY

• Main Author: Maulida Azka, Raka
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/78338
• Institution: Institut Teknologi Bandung

BUMI.JK is a stock issued by PT Bumi Resources Tbk. The stock is among the five stocks with the largest market capitalization for the coal sector. Public ownership with a share value below 5% is known to reach 75% of the total existing ownership. Therefore, investors affected by stock price movements will be more numerous and dominated by individual communities. This research aims to predict the stock price of BUMI.JK by using artificial intelligence at the deep learning level through the Long Short Term Memory (LSTM) method. This research also analyzes public sentiment towards BUMI.JK stock by predicting the sentiment that will be formed in the future. The results of self-learning by the LSTM algorithm will be validated with ARIMA which is a classical approach through statistical rules that can be understood by humans. Based on this research, stock price prediction with LSTM is validated by public sentiment analysis and ARIMA. It is known that the stock price has the largest possible increase of 66.8%, the expected increase of 26.1%, and the possibility of a decrease of 14.5% for the time range of June 2023 to June 2028. This research is expected to help investors in BUMI.JK shares to make decisions.