Perbandingan Kinerja Model Long Short Term Memory (LSTM) dan Facebook Prophet untuk Peramalan Harga Saham PT Aneka Tambang Tbk

Abstract

Stock price movements have complex and changing patterns and trends that become a major problem for investors and researchers in making predictions. Prediction and forecasting of stock prices with time series data has been done with various models. In previous studies, the LSTM and Prophet models have better advantages than other models. Therefore, in this study the authors compare the performance of the LSTM and Prophet models and perform forecasting using these two models which are expected to help in determining the model of choice in predicting and forecasting stock prices with time series data. The data used in this study are PT Aneka Tambang Tbk stock data from the beginning of 2019 to the middle of 2024. The data is divided into training data to train the model and testing data to evaluate its performance. Both LSTM and Prophet models are implemented with each library in the Python programming language. Model evaluation used in calculating accuracy is MSE, MAE, MAPE, and RMSE. The results showed that the LSTM model had better performance in predicting the share price of PT Aneka Tambang Tbk. The LSTM model gets an MSE value of 1387, MAE 28.72, MAPE 1.69%, and R2 0.96, while Prophet with an MSE value of 1960, MAE 34.34, MAPE 2.07%, and R2 0.96. Forecasting results using the LSTM and Prophet models show an upward and downward trend in stock prices 100 days ahead, the difference in the Prophet model is that the upward and downward trends tend to be stationary..