Predictive modelling on the effects of the critical parameters in grain storage systems: A case study in the Philippines

Abstract

Grain storage system plays a crucial role in preserving the quality and quantity of various grains including wheat kernels. Inefficient storage systems can result in spoilage, loss of quality, and reduction in quantity, leading to significant financial losses in the company. The quality of grains storage is influenced by several factors, including moisture content, protein content and falling number. In this study, the threshold, main, and interactions effects of critical parameters the identified critical parameters on grain storage quality were determined and modelled using ROC curve analysis and binary logistic regression analysis. The ROC curve analysis revealed a AUC of 0.992, 0.952, and 0.991 for moisture content, protein content, and falling number, respectively. Threshold values were also calculated and upon evaluation, the accuracies for all critical parameters yielded a significantly excellent result. The main effects of the critical parameters also significantly predicted the grain storage quality with ${\chi }^2$ of 353.26, p-value of 0.000, and $R^2$ value of 61.81%. Furthermore, the interaction effects of the three critical parameters exhibited the greatest effect on grain storage quality. Through binary logistic regression model, the optimum quality grain storage is achieved at a moisture level of 11.42, protein level of 13.99, and a falling number of 443.55. These findings can be used as basis for the selected milling company and other grain storage companies to monitor these critical parameters and ensure a safe storage and quality grains. The study can be furthered improved by investigating other parameters that can determine grain storage quality.