.... Our approach is similar to the one utilized by Colak et al. (2022) , which starts with 980 variables and ends up having the top 20 predictors in the final models for IPO failure risk prediction. In the second stage, we develop an IPO ranking framework using ML-derived feature importance...