Using machine learning techniques to study economic trends in various U.S. industries in the post-epidemic era

ABSTRACT

The aim of the project is to predict and analyse broad trends across the US economy using stock data from mainstream companies in six industries on Forbes 2000 and data from COVID-19. A time series analysis approach was used to predict the daily increases in each company's share price. The following five supervised learning techniques (logistic regression, random forest, decision tree, neural network and XGBoost) were used. As the accuracy of the results predicted by the different models for each company varies considerably, only the results predicted by the most accurate model for each company have been selected for analysed. The results show that the Electronic Pleased Technology Industry and the Social Entertainment Internet Industry remain break-even for COVID-19;the E-Commerce Industry shows a significant increase;The Financial Services Industry shows a significant drop in share price, while the Insurance Industry and Pharmaceutical Industry show a small drop in share price. © COPYRIGHT SPIE. Downloading of the is permitted for personal use only.