论文部分内容阅读
目的了解常用气温参数的多重共线性,为应用相关参数建立虫量的温度驱动消长模型提供参考。方法应用SPSS 13.0软件分析无锡市2010年3月至2011年11月每天的平均气温(X1)、最高气温(X2)、最低气温(X3)、18:00时气温(X4)、5日均温(X5)5个参数之间的相关性,并以本地蚊虫为例经向后法诊断了气温参数在蚊密度消长模型中的多重共线性。结果各参数之间呈直线相关,相关性均有统计学意义(P<0.01)。其中X1与X2的相关性最高,校正决定系数(Rc2)=0.959;X4与X5相关性最低,Rc2=0.811。在以气温参数为自变量建立蚊密度消长模型时,随着自变量个数的增加,多重共线性程度也增强,模型的复相关系数(R)升高,Rc2降低。结论气温参数之间存在显著的自相关,在建模时存在较强的多重共线性,多重共线性实际上降低了回归方程的拟合度,在引入气温因子建立相关模型时应避免同时应用多个气温参数。
Objective To understand the multicollinearity of commonly used temperature parameters and to provide a reference for establishing temperature-driven growth and decline models of insects using relevant parameters. Methods The average daily temperature (X1), maximum temperature (X2), minimum temperature (X3), temperature at 18:00 (X4) and average daily temperature at 5 ℃ of Wuxi from March 2010 to November 2011 were analyzed by SPSS 13.0 software. (X5) and the local mosquito was used as an example. The multivariate linearity of the temperature parameter in the mosquito density decrease model was diagnosed by the warp backward method. Results There was a linear correlation between the parameters, the correlation was statistically significant (P <0.01). Among them, X1 has the highest correlation with X2, the correction coefficient (Rc2) = 0.959; X4 has the lowest correlation with X5, and Rc2 = 0.811. When the model of mosquito density growth was established with temperature parameters as an independent variable, the degree of multicollinearity increased with the increase of the number of independent variables. The complex correlation coefficient (R) increased and Rc2 decreased. Conclusion There is a significant autocorrelation between temperature parameters, and there is a strong multicollinearity in modeling. Multicollinearity actually reduces the fitting degree of the regression equation. When establishing the correlation model by introducing temperature factors, it is necessary to avoid more simultaneous applications A temperature parameter.