Application of Factor, Cluster and Discriminant Analysis in the Evaluation of Style Characteristic of Tobacco Leaves

Pengze WANG,Pengfei LIU,Miao LAI,Peipei FU,Fuye ZHOU,Wei REN,Mingqin ZHAO

DOI: https://doi.org/10.13496/j.issn.1007-5119.2015.02.004

2015-01-01

Abstract:169 tobacco samples from 31 tobacco producing counties of Henan were carried out to study tobacco style and characteristics, aiming to provide a scientific basis for the rational evaluation of tobacco-style features and characteristics of different styles of tobacco area classification. Combined with factor analysis method, cluster analysis method and discriminant analysis method. Result: strong flavor type was little, with hay incense, burnt sweetness aroma and burnt aroma as the principal notes; the extreme significant positive correlation of flavor of Henan tobacco with the motion state of aroma, burnt-sweet aroma and burnt aroma of note and smoke consistence, the significant positive correlation with physiological strength; 5 main factors were extracted with factors analysis. 31 tobacco producing counties were divided into 3 categories through cluster analysis based on factors total score. Conclusion: fisher discriminant function evaluation model was established to judge clustering result, and the consistency of judge result and cluster analysis was 100.0%. The above results showed that this method applied in the evaluation of tobacco style features had strong feasibility and good effect.