Digital KPLSR model and multivariate quality dataset for storage assessment of achira-starch products

This dataset provides a comprehensive experimental runs and a Kernel Partial Least Squares Regression (KPLSR) digital modelling framework for assessing the multivariate quality evolution of achira-starch-based products during storage. The database includes the initial physicochemical and spectroscopic characterization of the products followed by a controlled 30-day storage study designed to monitor quality degradation. Measurements of moisture, water, colour, texture, and sensory attributes were collected at five time points (0, 5, 15, 20, and 30 days), enabling a detailed temporal analysis of product behaviour under standardized environmental conditions. The dataset is fully compatible with MATLAB® (The MathWorks Inc., Natick, MA, USA) and supports the development and calibration of predictive models for quality estimation using multivariate statistical tools, including KPLSR and data-driven empirical equations. These models allow simultaneous prediction of key quality indicators as functions of storage time, offering practical capabilities for shelf-life estimation, quality monitoring, and decision-making in product development and storage optimization.