Hybrid Machine Learning and Deep Learning Framework for Accurate Stock Price Forecasting

Abstract

Accurate stock price prediction and analysis are now essential for businesses and investors in today's market. It has been shown that the traditional methods for predicting and visualizing stock information are not accurate enough for capturing complex stock market trends and future data. This work proposes a novel hybrid approach that utilizes machine learning (ML) and deep learning (DL) techniques for intelligent stock price prediction and analysis. The main purpose of this study is to create a prediction model that can effectively investigate the historical stock data and accurately predict stock. The proposed model uses the Historical Stock Market Dataset having 9,909 records and 7 attributes, and data preprocessing, normalization, balancing using SMOTE class, along with suitable data splitting techniques are adopted to boost the reliability of prediction results. The Hybrid model presented in the paper is a combination of XGBoost+LSTM which extracts the nonlinear characteristics of both feature learning and temporal sequence in the stock market. The experimental results show that the proposed model has high predictive performance with an accuracy (ACC) of 93.9%, precision (PRE) of 93.7%, recall (REC) of 93.8%, and F1 score (F1) of 93.6%. The results demonstrate the feasibility of the proposed hybrid model in capturing complex relationships and sequential patterns in historical stock data, which has the potential to generate accurate predictions of stock prices and facilitate informed investment decisions.