论文部分内容阅读
预测是回归方程的一项重要的应用,评价一个经验回归方程的好坏,一个常用的重要标准就是看由它所作的预测的精度如何。因为在线性回归前提下,经验回归方程的形状,取决于选择的自变量,以及回归系数的估计方法。Hoerl和Kennard提出的岭回归(Ride Regression)与最小二乘法相比,其变量的选择,回归系数的估计及回归效果均令人满意。本文利用岭回归分析法从多元共线性的变量中,选择最佳的因子来建立水稻三化螟第二代螟害白穗率的预测方程,其结果要比多元回归的效果好。
Prediction is an important application of the regression equation. To evaluate the quality of an empirical regression equation, a commonly used and important criterion is how accurate the prediction made by it is. Because under the premise of linear regression, empirical shape of the regression equation depends on the choice of independent variables, and regression coefficient estimation method. Compared with the least-squares method, the choice of variables, the regression coefficient estimation and the regression effect of Hoerl and Kennard’s ridge regression are all satisfactory. In this paper, we use ridge regression analysis to select the best factor from multivariate collinearity variables to establish the prediction equation of white spike rate of the second generation rice stem borer, Chilo suppressalis. The result is better than the multiple regression.