Explainable hybrid quantum neural networks for analyzing the influence of tweets on stock price prediction

摘要: Stock price prediction is a complex and challenging activity for organizations and investors to predict future returns. While machine learning and deep learning methods are widely used for stock closing price prediction, these methods have some drawbacks, including high scalability, slow convergence, and poor generalization performance. Furthermore, because those models are inherently black-box, it is challenging to comprehend the logic underlying their forecasts. This study presents an explainable hybrid quantum neural network to investigate the influence of tweets on a stock price prediction. The datasets used in this analysis include the stock prices of six different organizations as well as the 4 million+ tweets written on X (previously Twitter). The proposed methodology finds the average sentiment score of daily tweets using a transformer model which is combined with historical stock data. The proposed …