Comparative Study on Stock Price Forecasting Using Deep Learning Method Based on Combination Dataset

Stock forecasting is the process of employing various analysis methods and mathematical models, including deep learning techniques, to predict future stock price movements based on historical data and relevant market factors. This paper aims to contribute to the field of stock price prediction by introducing a comprehensive forecasting model. The model integrates OHLCV, technical indicators, macroeconomic variables, and fundamental dataset, leveraging a multifaceted dataset approach. Through the incorporation of these diverse datasets, the proposed model seeks to enhance the accuracy and robustness of stock price forecasts, providing a more holistic understanding of market dynamics for investors and researchers alike. In conclusion, the test results indicate that the application of a combined dataset using feature selection, along with the utilization of the TFGRU model, yielded positive results. The model achieved an RMSE of 16.19, a MAPE of 2.65, and an AcMAPE of 0.85. Lower RMSE and MAPE values suggest enhanced performance, and the relatively low AcMAPE, considering both accuracy and percentage error, further underscores a favorable outcome.