论文部分内容阅读
以农业部测土配方施肥项目2006～2007年在辽宁省进行的34个花生试验数据为依据,比较向后回归法建立肥料效应回归模型与强迫法建立肥料效应回归模型的优劣,探讨对关键处理设置重复在建立数学模型中起到的作用。结果表明,采用强迫法进行三元二次方程拟合的成功率只有12%,而用向后回归法拟合的成功率有55%,2007年对处理1、处理2、处理4、处理6、处理8设置重复后,三元二次方程模型和一元二次方程模型方程的拟合的成功率分别占当年试验总数的84%、16%。2006年不设置重复,有20%不能够拟合。因此,“3414”肥料试验关键处理设置重复,并采用向后回归法,对“3414”肥料试验得到合理施肥模型具有重要作用。
Based on the data of 34 peanut experiments carried out by the Ministry of Agriculture in 2006 ~ 2007 in Liaoning Province, the authors compared the regression model of fertilization effect with the forcing method to establish the regression model of fertilization effect, Process Settings Duplicate The role that plays in establishing a mathematical model. The results show that the success rate of compulsory triple quadratic equation fitting is only 12%, while the regression rate of 55%, the success rate of 2007, the treatment of 1, treatment 2, treatment 4, treatment 6 , The success rate of the fitting of the ternary quadratic equation model and the quadratic equation model equation after the repetition of treatment 8 is set to be 84% and 16% of the total number of the same year respectively. No duplication was set in 2006, and 20% could not be fitted. Therefore, the key process of “3414” fertilizers test was repeated and backward regression method was used, which played an important role in obtaining reasonable fertilization model in “3414” fertilizer experiment.