作物品种区域试验中品种均值的Bayes估计

Bayesian Estimation of Variety Means in Regional Crop Trials

本研究根据 Bayes统计原理 ,提出作物品种区域试验中品种均值的 Bayes估计方法 ;并利用一套包含 4年、7个试点和 10个棉花品种的多年多点试验对 Bayes估值和算术平均值的预测精度进行比较。结果表明 ,Bayes估值和算术平均值的平均预测差分别为 6.88%和 12 .77% ;而且 Bayes估值与验证值之间在数值和品种排序上都有着更高的相关 ,相关系数分别达 0 .963和 0 .976,高于算术平均值的 0 .876和 0 .830 ;

Based on the principle of Bayesian Statistics, a Bayesian method for estimating variety means in regional crop trials was proposed. A trial involving 4 years and 7 sites and 10 varieties of cotton was carried out and analyzed to compare the predictive accuracy of the Bayesian estimates and arithmetic means. It was found that the averaged predictive differences of Bayesian estimates and arithmetic means were 6.88% and 12.77% respectively; the coefficients of correlation between Bayesian estimates and validation data were 0.963 for values and 0.976 for variety ranks, higher than 0.876 and 0.830 of between arithmetic means and validation data. Bayesian estimates showed higher predictive accuracy than arithmetic means.