期刊
INTERNATIONAL DAIRY JOURNAL
卷 15, 期 6-9, 页码 557-562出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.idairyj.2004.10.010
关键词
artificial neural network; discriminant analysis; Emmental cheese; authenticity; chemometric data analysis
The goal of this work was to classify European Emmental cheeses according to their geographic origin using analytical approaches. Twenty-five analytical parameters (factors) were measured in 183 samples. Results were combined by multivariate statistical analysis. Discriminant analysis (DA) and an artificial neural network (ANN) delivered similar results when all regions and factors were included, 95 % and 91 %, respectively, of the samples were correctly classified in the validation procedure. To reduce the analytical costs and the risk of overfitting, a DA based on a selection of only 11 factors was calculated. In this case, the Jackknifed validation delivered 95 % correct assignments. Finally, a system was optimised to discriminate between the Swiss samples and cheeses from other regions. Building a new model for each of the six pairs, Switzerland vs. another region, 100 % correct classification could be achieved for the Swiss samples. (c) 2005 Elsevier Ltd. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据