Combined Experimental and Predictive Modeling of Jasmine Flower Shelf Life Using Machine Learning

This study investigates the impact of packaging materials and cold storage on the postharvest quality and shelf life of jasmine (Jasminum multiflorum) flowers and applies machine learning algorithms to predict shelf life based on key quality indicators—physiological loss in weight (PLW), browning index (BI), and total phenol content. Flowers were stored at 5°C ± 2°C for 20 days in three packaging types: CFB box with polyethylene (PE) lining, bamboo basket with newspaper lining, and nylon bag. Maximum shelf life was observed in CFB with PE lining (16 days) and minimum in bamboo basket (8 days). Machine learning models—Random Forest (RF), Bayesian Regularized Neural Network (BRNN), and Support Vector Machine (SVM)—were trained using the experimental data. Among them, BRNN achieved the best performance with RMSE = 1.03, MSE = 1.06, and R 2 = 0.97 for shelf‐life prediction, outperforming RF (RMSE = 1.96, R 2 = 0.91) and SVM (RMSE = 1.46, R 2 = 0.94). The study demonstrates the potential of integrating physiological data with machine learning models for accurate shelf‐life prediction and quality management in floriculture. This study uniquely combines physical experimentation with AI‐based prediction, offering a dual contribution of empirical validation and scalable digital forecasting for jasmine flower shelf life.