The Trend of Published Research for Predicting Bankruptcy Via Machine Learning: A Bibliometric Analysis

Bibliometric analysis is a valuable technique for quantitative research on academic output to analyze pre-existing materials and address research patterns in a particular field. This paper aims to investigate machine learning approaches in predicting business failure. The findings suggest that author collaboration remains lacking despite a recent rise in publications, particularly internationally. Furthermore, We note that the subject of bankruptcy prediction has not placed enough emphasis on some machine-learning techniques. For example, Convolutional Neural Networks (CNNs) excel at processing images rather than numbers. In addition to (CNN), Long-Short Term Memory (LSTM) networks have also faced limitations in their application within this field due to several factors, including the requirement for large datasets and the necessity to investigate the “black box” of these models. These obstacles present opportunities for researchers to improve existing work in this area. This study has implications for theory and practice. In the first place, it gives academics knowledge about the intellectual framework and theoretical development of this topic, which is helpful for future research. Secondly, the study illuminates underutilized machine learning techniques for bankruptcy prediction, which hold significant value for corporate firm managers and policymakers regarding management and decision-making.