A COMPARATIVE ANALYSIS OF MACHINE LEARNING ALGORITHMS FOR IPO UNDERPERFORMANCE PREDICTION

Abstract

Initial Public Offerings (IPOs)  are a popular way for companies to raise capital and enter the public markets. However, many IPOs underperform and fail to meet the expectations of investors. In this research paper, we explore the use of different machine learning models, namely AdaBoost, Random Forest, Logistic Regression, ANN and SVM, for predicting IPO underperformance. We collect and pre-process a dataset of IPOs from the past few years, and use it to train and evaluate the performance of each model. Our results show that Artificial Neural Network model is better suited for predicting IPO underperformance. Additionally, our analysis provides insights into the factors that contribute to underperformance and highlights the importance of certain features in predicting IPO performance. Our research provides valuable information for investors and financial analysts interested in predicting the performance of IPOs and mitigating the risks associated with IPO investments. We have tested machine learning models, namely AdaBoost, Random Forest, Logistic Regression, ANN and SVM. After Comparing the accuracy of all the models, we arrived at the conclusion that ANN model performed the best with an accuracy of 68.11%.