Machine learning-based process optimization and quality prediction for hot-air drying of mulberry

This dataset supports the study "Machine learning-based process optimization and quality prediction for hot-air drying of mulberry". A three-factor, three-level L9(3^4) orthogonal experiment was conducted to investigate the effects of drying temperature (55, 60, and 65 °C), drying time (6, 8, and 10 h), and air velocity (1.0, 1.5, and 2.0 m/s) on six quality attributes of mulberry (Morus alba L.): water loss rate, hardness, springiness, chewiness, total soluble solids (TSS), and titratable acidity (TA). The dataset comprises 45 samples (9 experimental groups × 5 replicates) and includes the orthogonal experimental design with measured responses, organized model-training tables, and a trained random forest regression model (with standardization and polynomial feature expansion, optimized by 5-fold cross-validation, R² = 0.960–0.988) together with the identified optimal drying conditions (65 °C, 10 h, 1.5 m/s). Files are provided as Microsoft Excel workbooks (.xlsx) and a JSON model file.