Establishment of the Optimal Model on Chemical Content in Flue-Cured Tobacco and Smoke Composition

Abstract:

Regression analysis and statistical methods were used to establish the optimal model of the chemical content and smoke composition.The results show that:1) equations were examined in full model,double screening stepwise regression,and stepwise regression only containing relevant variable.However,the prediction accuracy of the model was different in these methods,and not all the regression coefficient of the predictor variables was examined.2) when the indistinctive variable removed,the higher prediction accuracy and regression diagnostics models about Tar,nicotinamide and CO were obtainted.Tar regression equation was Ytar=11.7+1.857Xnicotine,Nicotine regression equation was Ynicotion=0.679Xnicotion,CO regression equation was YCO=1.428 6Xnicotion+0.121 7Xreducing sugar+2.822 5Xtotal chlorine+8.488 9,and all the regression equation test were significant(P<0.01).