AI-driven stock market prediction

Abstract

The accuracy of deep learning techniques used for prediction has always been deemed superior as compared to regression techniques. In this report, deep learning techniques such as Long Short-Term Memory, Recurrent Neural Network, Multi-Layer Perceptron and Gated Recurrent Unit will be used in a comparison with regression techniques such as Gradient Boosting Regressor and Support Vector Regressor to forecast the Straits Times Index (STI). The data sourced will also be non-linear and will be used as inputs into the algorithms to generate the results. The results will be compared using Fundamental Analysis and Technical Analysis. This experiment shows that the results from deep learning techniques does not generally mean that it is more accurate as compared to regression techniques.