Prediction of stock closing price PT. Telkom Indonesia with LSTM method

• Main Authors: Prasetyo, Firman Aji, Nugroho, Setiya, Setiawan, Agus
• Format: Conference Proceeding
• Language: English
• Subjects: Investments; Model accuracy
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0122363

Stock investment is an investment activity that is currently in great demand by the public. In fact, there are still many investors who are doubtful about the risks of investing, worrying that what they are investing in does not meet their expectations. The doubts of potential investors in investing are caused by fluctuations in the stock price index within a certain period of time. For this reason, so that investors do not lose when buying stock, it is necessary to analyze the stock to be purchased. Then this study will focus on the discussion of making a model that can be used to predict closing stock prices name "TLKM" using Long Short Term Memory Method. LSTM stores pattern information in the data and can manage memory at each input using memory cells and gate units, so it can combine current information with past information, making it very efficient to record old information. While the results of the LSTM model training with train data resulted in loss with MAE and using the ADAM optimizer which resulted in a low value of 0.0178. Then testing the data to make model predictions, which results in an error rate using MAPE which is 1.62% which is based on the MAPE calculation criteria, the calculation is in the very good category. Then for the level of accuracy of the model that has been made using R squared, the resulting accuracy rate is 95.78% which is fairly good, approaching 100%.