主成分分析和聚类分析在烟叶全面质量评价中的应用

The Application of the Principal Component Analysis and Cluster Analysis in Overall Quality Evaluation of Tobacco

为研究烟叶全面质量评价的新方法,2014年于四川省宁南县采集烟叶样品,测定分析了烤烟的外观质量、物理特性、化学成分、中性致香物质和感官质量,并应用主成分分析和聚类分析对烟叶38个质量指标进行了分析和聚类。结果表明,38个指标共提取出了7个主成分,方差累计贡献率为87.37%,基本上反映了原来变量的全部信息;烟叶样品经过聚类分析共分成3类,经过判别分析验证,正确率达92.86%;最适宜种植烤烟的海拔高度为1 450 m,其次为海拔1 750 m,海拔2 050 m种植效果最差,黄壤种植效果优于红壤,云烟87和云烟85种植效果优于云烟97。

In order to study the new method of overall quality evaluation of tobacco, tobacco leaves were collected from Ningnan county of Sichuan province in 2014, and their appearance quality, physical properties,chemical components, neutral aroma components and sensory quality were determined and analyzed. 38 quality indexes of tobacco leaf were analyzed and clustered by principal component analysis and cluster analysis. The results showed that,seven principal components were extracted from 38 indexes, and cumulative contribution rate of variance was 87.37%,which basically reflected all information of original variable; tobacco leaves were divided into three categories by cluster analysis,and correct rate reached 92.86% by discriminant analysis. The optimum altitute for planting flue-cured tobacco was 1 450 m,followed by the altitute 1 750 m, and2 050 m in turn, the planting effects of yellow soil was better than that of red soil,and the planting effects of Yunyan 87 and Yunyan 85 were better than that of Yunyan 97.