This study develops a multi-objective prediction model to solve complex prediction tasks in hierarchical data structures. The first is the random forest model, which improves the accuracy and stability of the model by constructing multiple decision trees and combining their predictions while solving the nonlinear dependency and convergence problems. The random forest model efficiently models complex relationships through global optimization of initial weights and biases. The second approach is an XGBoost model that utilizes advanced feature construction techniques focusing on improved feature tuning and regularization techniques to achieve a balance between accurate error correction and complex pattern capture. The framework emphasizes the importance of feature engineering, integrating objective and subjective feature weighting to improve the accuracy of multivariate datasets. By fusing machine learning methods with statistical paradigms, this integrated model improves predictive performance and provides actionable insights for complex and diverse use cases.
Liu et al. (Tue,) studied this question.