Innovative Design Path of Hidden Interactive Structure in Water Packaging: Application of Machine Learning Models

Packaging has evolved from a purely functional aspect to a strategic vehicle (a useful business tool) that conveys meaning, evokes emotions, and inspires action. This paper examines the impact of interactivity, usability, design, and sustainability on consumer relationships with packaging, repeat purchasing, and social sharing behaviors. Based on survey data, three machine learning (ML) models — Random Forest, XGBoost, and Logistic Regression — were used to predict outcomes in behavior and assess feature importance. Results show that XGBoost consistently outperformed all other models, achieving an F1-score of 0.9655 and an AUC of 0.9881 for repeat purchase, and remained a perfect predictor for social sharing, with F1 = 1.0000 and AUC = 1.0000. The strong predictor was interactivity, followed directly by usability, willingness to co-create, and emotional connection. This highlights the market opportunity that experiential packaging offers to foster strategic consumer interaction and loyalty for a brand, while also conveying a message of cultural significance and sustainable material innovation. The use of survey data and the narrow demographic of the study suggest opportunities for future work in large and cross-cultural samples, with sustainable and consumer-centered packaging design that can be validated with real-world experimental methods.