An Advanced Prediction Model for Risk Assessment of Imported Food

Previous studies have primarily concentrated on the development of systems designed to predict and manage the risk associated with imported foods. However, to achieve more accurate inspection results, it is essential to enhance the performance of these prediction models. Hence, this study aims to propose methods for improving the performance of risk prediction models for imported foods. Through a series of model enhancement experiments, we have confirmed that techniques such as item name risk derivation, feature generation, dimensionality reduction, and stacking ensemble significantly contribute to model performance improvement. The findings of this study are expected to provide a strategic direction for more effective management of imported food safety and to serve as a valuable resource for future research.