摘要
利用 1982年以来我国棉花、小麦、水稻和玉米的 60套区域试验数据 ,采用交叉验证方法 ,对区域试验中算术平均值、最佳线性无偏预测值 ( best linear unbiased predictor,BLUP)和 AMMI( additivemain effects and multiplicative interaction)模型估值的预测精度进行比较 ,结果表明 ,与算术平均值相比 ,AMMI估值精度的增益倍数 ( gain factor,GF)平均为 1.0 4 5,变幅为 0 .963~ 1.4 14,其精度多数情况下提高不大 ;BLUP的 GF平均为 1.170 ,变幅为 1.0 0 8~ 1.619,其精度普遍较高。同时 ,文中对
Sixty sets of data from regional trials of cotton, wheat, rice and maize since 1982 in China were used in cross validation to compare the predictive accuracy of arithmetic means and BLUPs( best linear unbiased predictors ) and AMMI ( additive main effects and multiplicative interaction) estimates. The average precision gain factor(GF) of AMMI relative to arithmetic mean was 1.045 with a range from 0.963 to 1.414, which showed slight increases of precision; BLUP was found commonly to outperform arithmetic mean and AMMI with an average GF of 1.17, ranged from 1.008 to 1.619.
出处
《作物学报》
CAS
CSCD
北大核心
2001年第4期428-433,共6页
Acta Agronomica Sinica
基金
国家自然科学基金资助项目 (30 0 70 433)