The impact of social media and the pandemic on stock price prediction

Predicting stock prices has been a widely studied topic across numerous disciplines for a long time. In this study, the goals are to first, analyze how social media sentiment influences stock price predictions; second, compare the effects of social media sentiment on stock price predictions before and during the pandemic; third, investigate the impact of the pandemic on stock prices across three major sectors: airlines, hotels, and restaurants. This research leverages three distinct types of data – stock prices, COVID-19 data, and social media data – to develop three separate feature sets for analyzing the impact of various factors on RNN-based stock prediction. The process begins with loading the relevant datasets and initializing a sequential model. Next, the model is built by adding an input layer, followed by an LSTM layer, and one or more dense layers. After the model is compiled and trained, it is evaluated, and the results are visualized to assess the outcomes. After analyzing the research results, we find that: (1) Social media sentiment affects stock price prediction in all three sectors; (2) Social media sentiment is a more effective predictor of stock price trends across all three sectors during the pandemic compared to the pre-pandemic period; (3) Although the impact of COVID case data on stock price prediction is not consistently observed across all three sectors, it significantly enhances model performance in several instances. Notable improvements are seen in specific stocks, indicating that pandemic-related data may have a more substantial effect on certain companies within the sectors. These findings offer valuable insights into recovery strategies and investment opportunities within each industry.