初烤烟叶霉变率随时间变化的SPSS曲线拟合分析  

Curve Estimation of Dynamics Changes between Rate of Flue-Cured Tobacco Leaves Spoilage and Storage Time by SPSS

作　　者：袁逢春[1] 苏罗毅 李庆华[1] 胡伟[1] 飞旭云 李文鹏[2] 

机构地区：[1]红云红河烟草(集团)有限责任公司昆明卷烟厂,云南昆明650202 [2]云南大学生命科学学院,云南昆明650091

出　　处：《西南农业学报》2014年第3期1130-1135,共6页Southwest China Journal of Agricultural Sciences

基　　金：红云红河烟草(集团)有限责任公司科技项目(HYHH2010HX03)

摘　　要：本研究旨在筛选一个合适的模型,以表征初烤烟叶霉变率随时间变化的规律。为达到研究目的,以昆明卷烟厂整理车间2007-2010年3个烤季在不同时间段的烟叶霉变率平均值为基础,利用SPSS软件进行曲线拟合,采用偏差因子(Bf)和准确因子(Af)评价预测模型的可靠性。结果表明,初烤烟叶霉变率与时间两个变量之间呈极显著正相关(P<0.01),在7种非线性模型中,Cubic(三次曲线)模型的R2值均大于其他模型,而且只有Cubic模型的判定系数R2值(0.999)大于0.99,非常接近1,标准误最小(0.031),因此Cubic模型最合适用来表征初烤烟叶霉变率随时间变化的规律。Cubic模型的三个烤季平均偏差因子为1.04,平均准确因子为1.28,故通过验证,Cubic模型能快速准确地预测初烤烟叶随时间变化的霉变率。This study aimed to explore a suitable mode characterizing dynamics changes of the rate of flue-cured tobacco leaves （FCTL） spoilage over the storage time.In order to achieve the objective of this study,nonlinear curve fitting was performed with curve estimation by SPSS based on average rate of FCTL spoilage in different storage time of 2007-2010 from Kunming Cigarette Factory.Model performance was evaluated with predicted versus observed values using the prediction bias （Bf） and accuracy factors （Af）.The results showed that there was an obviously positive correlation between the rate of spoilage and the storage time （P 〈0.01）.Comparison of predicted versus observed values of rate of FCTL spoilage indicated that the cubic model fits better than other models （quadratic,compound,growth,exponential,logistic and power）,with regression coefficients R2 〉 0 99 and the standard error =0.031.The Bf and Af for the cubic model were 1.04 and 1.28,respectively.Therefore,the cubic model can accurately predict the change of the rate of FCTL spoilage according to storage time.

关 键 词：初烤烟叶 霉变率 SPSS 曲线估计 偏差因子 准确因子 

分 类 号：TS445[农业科学—烟草工业]