Systematic Literature Review

Stock Price Prediction Using Machine Learning and Deep Learning

Abstract

In order to determine the type of evaluation with the attributes utilised, the methods employed, the methods most frequently used, and the methods with the greatest performance, this research was carried out using a literature review method to analyse various studies. Based on pre-established criteria, 40 publications were selected from research conducted between 2016 and July 2021 for this study. Four research topics estimation, classification, clustering, and association—were determined by this review. Four research topics fundamentals, technical, sentiment analysis, and even a combination with analyses that make use of their respective qualities and datasets—are the results of this study. Thirty-one different methods have been identified for stock price prediction. The most popular approaches are LSTM, MLP, RF, and SVM. Furthermore, MLP is a technique that yields the best results, with an LSTM of 70% and a performance of 71.63%. For better precision, it is advised to employ deep learning, ensemble approaches, mixed machine learning, and input attribute selection at pre-processing.