Stock Market Trend Analysis and Machine Learning-based Predictive Evaluation

Abstract

Financial experts may make successful selections thanks to the stock market's research and forecasting capabilities, which is exciting. This study examines the stock market forecast outcomes through a simple feed-forward neural network (FFNN) model. Then, we contrast those outcomes with those produced using more sophisticated Elman, fuzzy logic, and radial basis function networks. Any problem with finite input-output mapping may be solved using the FFNN as long as it has at least one hidden layer and a sufficient number of neurons. An ANN in which RBFs are used as activation functions is called a radial basis function network (RBFN). Utilizing the Levenberg-Marquardt Back Propagation technique, the FFNN and Elman networks are trained in this study. A Fuzzy Inference System (FIS) of Sugeno type is employed to replicate the predictive procedure within the realm of fuzzy logic. We choose the optimal RBF values using several clustering techniques. The approaches were validated using public stock market data on the National Stock Exchange of Indonesia.