Abstract

Prediction and classification of the stock price movement remain crucial to traders and financial analysts. The complexity and fluctuations of the stock market make its prediction and classification a difficult task. In this study, we compared the performance of the machine learning and deep learning models in predicting and classifying the Amazon online stock market Index. The prediction and classification have resulted in high accuracy compared to previous studies. After applying the single-step wrapping method, the second stage resulted in the best six features. The sequential ANN with hard-sigmoid transfer function and Deep learning with Rectified Linear Unit (ReLU) transfer function has correctly predicted and classified the price movements. The experimental accuracy of the prediction models has improved by more than 15% compared to the previous studies. The proposed model may be applied to other stock market indices to evaluate their price movement.

Author: Godfrey Joseph Saqware and Ismail B.

Received on: January, 2022

Accepted on: March, 2023