Long Short-Term Memory Algorithm for Stock Price Prediction

Stock prices in the capital market fluctuate from time to time, many factors influence it. Investors need to do an accurate analysis to reduce the risk of investing, one of which is by predicting stock prices. The results of the predictions help investors to make decisions. The right decision requires accurate prediction results. So it is necessary to predict stock prices so that investors can understand investment prospects in the future. In this study, the LSTM algorithm will be used. The LSTM algorithm can extract information from long-term, time series or sequential data. The resulting MAPE value of 2.2% of these results is in the very good category because it is less than 10% and the resulting R2 of 0.974 is close to the value of 1. So that stock predictions using LSTM are included in the category of very good stock prediction models. Produce optimal stock predictions on the comparison of training data and testing data of 70:30 with 500 epochs and 64 batch sizes.