Artificial Intelligence and Exploratory-Data-Analysis-Based Initial Public Offering Gain Prediction for Public Investors

Abstract

An initial public offering (IPO) refers to a process by which private corporations offer their shares in a public stock market for investment by public investors. This listing of private corporations in the stock market leads to the easy generation and exchange of capital between private corporations and public investors. Investing in a company’s shares is accompanied by careful consideration and study of the company’s public image, financial policies, and position in the financial market. The stock market is highly volatile and susceptible to changes in the political and socioeconomic environment. Therefore, the prediction of a company’s IPO performance in the stock market is an important study area for researchers. However, there are several challenges in this path, such as the fragile nature of the stock market, the irregularity of data, and the influence of external factors on the IPO performance. Researchers over the years have proposed various artificial intelligence (AI)-based solutions for predicting IPO performance. However, they have some lacunae in terms of the inadequate data size, data irregularity, and lower prediction accuracy. Motivated by the aforementioned issues, we proposed an analytical model for predicting IPO gains or losses by incorporating regression-based AI models. We also performed a detailed exploratory data analysis (EDA) on a standard IPO dataset to identify useful inferences and trends. The XGBoost Regressor showed the maximum prediction accuracy for the current IPO gains, i.e., 91.95%.