Predictive Modelling of the Bread-making Performance and Dough Properties of Wheat

Abstract

Data describing the basic chemical composition, Farinogram, Extensogram and test baking results from 100 wheat samples were analysed by principal component analysis and partial least-square regression. The stabilities of variables in models for prediction of bread volume were investigated. Variables (e.g. bulk density, flour yield and Falling Number) with large differences in correlations with bread volume in different sample groups (cultivar and year of harvest) had to be excluded in order to create stable models. Variables related to protein content were identified as stable, and could be combined in a global predictive model. Data describing the content and composition of non-starch polysaccharides were available for 49 samples, but inclusion of these variables did not improve the performance of the model for predicting bread volume. Farinograph water absorption was largely explained by non-starch polysaccharide content, composition and structure, however.