Predicting Gen Z Investment Perception Through Digital Financial Literacy and Ipo Awareness: an Interpretable Ai Framework

Abstract

This study develops an interpretable artificial intelligence framework for predicting Investment Perception among Generation Z using Digital Financial Literacy, IPO Awareness, Technology Innovativeness, Trust, Financial Attitude, and Digital Financial Self-Efficacy. A secondary survey dataset comprising 426 respondents was analyzed through comparative machine learning, hyperparameter optimization, stacking ensemble modelling, SHAP-based interpretation, repeated cross-validation, bootstrap uncertainty assessment, and ablation analysis. Five baseline algorithms, including Logistic Regression, SVM-RBF, Random Forest, XGBoost, and Gradient Boosting, were evaluated before the optimized learners were incorporated into a stacking ensemble. The proposed model achieved cross-validated accuracy of 0.707, F1-score of 0.780, ROC-AUC of 0.749, and MCC of 0.397. On the held-out test set, accuracy was 0.640, F1-score was 0.726, and ROC-AUC was 0.686. SHAP analysis identified Digital Financial Self-Efficacy as the strongest global predictor, followed by Trust and IPO Awareness. IPO Awareness accounted for 15.06% of global SHAP importance and demonstrated positive incremental predictive value in the ablation analysis. Digital Financial Literacy remained relevant to individual predictions but showed comparatively limited unique incremental contribution. Overall, the findings demonstrate the value of interpretable ensemble modelling for understanding Generation Z investment perception and provide transparent, data-driven insights for financial educators, investment platforms, and fintech stakeholders.