This study investigates the effects of ethanol treatment on kudzu starch gels using microstructural analysis, near-infrared (NIR) spectroscopy, and advanced data analytics to develop a nondestructive method for predicting gel strength and classifying treatment levels. Scanning electron microscopy revealed network densification with increasing ethanol concentration. NIR spectroscopy, combined with various variable selection methods and modeling algorithms, developed predictive models for gel strength. The Uninformative Variable Elimination - Support Vector Machine (UVE-SVM) model performed exceptionally with the highest R