Amazon Stock Price Prediction During COVID-19 Based on the LSTM Model and Linear Regression Model

Abstract

With COVID-19 sweeping the world in 2019 and the ensuing turmoil in the stock market, the need for accurate stock forecasting becomes particularly crucial during exceptional periods. Utilizing Amazon stock price data from 20022023, this study uses a machine learning approach to create a stock price prediction model in order to more accurately anticipate the trend of the stock price movement during COVID-19. The linear regression model and LSTM model are two models commonly used in forecasting studies, and this paper uses both models to forecast stock prices separately and evaluates the models by RMSE. The linear regression model is constructed directly with all data as independent variables. According to a 9:1 ratio, the LSTM model is segmented into a training set and a test set in this paper, with an overall structure of three layers and an optimizer of Adam optimizer. The test results reveal that the LSTM model's error is significantly lower than the linear regression model's error, proving that the LSTM model has a superior forecasting capacity to that of the linear regression model. As a result, it can be widely applied to the task of predicting stock prices for COVID-19 and offering decision-making advice to the appropriate decision-makers.