摘要
以评吸人员对12个城市进行评吸收集后得到的3900个数据为材料,分析了4类特性11个指标,其中包括香气特性(香气韵调,香气质,香气量,谐调性等4个指标)、烟气特性(浓度,劲头等2个指标)、口感特性(刺激性,湿润度,干净度,回味等4个指标)、杂气特性(即杂气指标)之间的典型相关分析以及4类评吸特性中11个指标之间的简单相关分析。结果表明:1.由简单相关分析看出11个评吸指标间均存在极显著的线性正相关关系,并且在香气韵调、香气质、香气量3个指标之间,香气韵调与杂气,湿润度与干净度存在高度正相关,其余指标间只存在低度和中度正相关。2.由典型相关分析和简单相关分析综合反映的香气特性与烟气特性的相关性主要可由谐调性、香气量与浓度间相关代表,烟气特性与口感特性的相关性主要可由浓度与湿润度及回味间相关代表,烟气特性与杂气特性的相关性主要可由浓度与杂气间相关所代表,香气特性与杂气特性的相关性主要可由香气韵调与杂气间相关代表,口感特性与杂气特性的相关性主要可由干净度、湿润度及回味与杂气间相关代表。3.由于香气特性与口感特性间多指标综合影响,所以利用香气特性第1典型变量得分来预测口感特性第1典型变量得分的回归方程来表示,且从散点图上也可观察到显著的线性关系。
Correlation analysis of smoking indicator characters for tobacco products was carried out using 3900 data collected from 12 cities. Four characters of 11 indicators, i.e. aroma characteristics （aroma rhyme, aroma quality, aroma volume, harmonization）, cigarette smoke characteristics （taste strength, physiological strength）, taste characteristics （irritation, moisture, cleanness, remaining taste）, offensive odor characteristics （offensive odor index） were studied and simple correlation analysis of the 11 indicators regarding these characters were conducted. Results showed that： 1. there was significantly positive linear correlation among 11 indicators and a high degree of correlation among aroma rhyme, aroma quality, and aroma volume, aroma rhyme and offensive odor, moisture and cleanness. There was only low and moderate positive correlation among other indicators. 2. From canonical correlation analysis and simple correlation analysis, the correlation between aroma characteristics and smoke characteristics can be represented by the correlation between harmonization, aroma volume and taste strength; the corre-lation between smoke characteristics and taste characteristics can be represented by the correlation between taste strength and moisture, remaining taste ; the correlation between smoke characteristics and offensive odor characteristics can be represented by the correlation between taste strength and offensive odor; the correlation between aroma characteristics and offensive odor characteristics can be represented by the correlation between aroma rhyme and offensive odor; and the correlation between taste characteristics and offensive odor characteristics can be represented by the correlation between cleanness, moisture, remaining taste and offensive odor. 3. Because aroma characteristics and taste characteristics are combined effects of multiple indicator, so the regression equation which uses the aroma characteristics＇ first canonical variable scores to predict the taste characteristics＇ first canonical variable scores could express the correlation between aroma characteristics and taste characteristics. From the scatter plot also observed a significant linear relationship.
出处
《中国烟草学报》
EI
CAS
CSCD
2008年第6期13-18,共6页
Acta Tabacaria Sinica
基金
云南中烟工业公司科技项目“区域因素对卷烟配方设计的影响及相关性分析研究”(合同号:2006JC02)
关键词
典型相关分析
简单相关分析
回归分析
卷烟产品
评吸指标
canonical correlation analysis
simple correlation analysis
regression analysis
cigarettes sampled product
smoking indicators
