作物学报 ACTA AGRONOMICA SINICA 2016, 42(1): 141-148 http://zwxb.chinacrops.org/
ISSN 0496-3490; CODEN TSHPA9
E-mail: xbzw@chinajournal.net.cn DOI: 10.3724/SP.J.1006.2016.00141
一种新的符合度算法及其应用
张 慧1,2 顾世梁1,* 李 韬1
1
扬州大学农学院, 江苏扬州225009; 2桐庐县农业技术推广中心, 浙江杭州311500
摘 要: 在总结分析了几种常用综合评价方法的基础上, 提出了一种反映观察值与理论值之间相似性的新算法——符合度。该算法就评价信息个体(观察值)与标准值(期望值)的马氏距离, 再由马氏距离转化为评价对象与标准的接近程度, 即符合度(r)。首先进行指标数(p)、相似度(r)与马氏距离(d)的模拟试验, 再通过曲面拟合的方法找出它们之间的关系模型。通过大量抽样试验, 验证符合度的次数分布与原先设定的符合度的良好对应关系, 说明模型的可行性与可靠性。以小麦RVA性状为指标, 利用该算法分析扬麦系统若干品种之间的接近程度, 并评价多变数复杂效应回归分析模拟试验的结果。符合度算法不需要数据标准化处理, 直接利用原始数据, 减少了计算工作量, 降低了因数据标准化处理方法不同而引起的评价结果差异, 同时由于不需要赋权, 排除了主观性的影响, 保证了信息的完整性以及评价结果的可靠性。
关键词: 符合度; 综合评价; 计算机模拟; 马氏距离
A New Algorithm for Conformity and Its Application
ZHANG Hui 1,2, GU Shi-Liang 1,*, and LI Tao 1
1
Agricultural College of Yangzhou University, Yangzhou 225009, China; 2Agricultural Extension Station of Tonglu County, Hangzhou 311500, China
Abstract: This article proposed a new algorithm of conformity using original data to calculate similarities between the target ob-ject and the expected value based on the Mahalanobis distance, providing an objective and reasonable analysis. Firstly, simulation experiments were conducted to obtain Mahalanobis distance (d) related to number (p) of different variables (traits) and similarity (r). Then, a surface fitting method was used to establish the function relationship between conformity (r) and index number (p), as well as Mahalanobis distance (d). Monte Carlo experiment for frequency distribution of conformity verified its good performance of the relationship model. The simulation results fully validated the feasibility and reliability of the model. Conformity algorithm was applied to calculate the similarity of a panel of Yangmai wheat varieties released in recent years referring to RVA parameters. The assessment of simulated multivariate regression for complex effects was also conducted. This study showed that conformity algorithm using raw data directly instead of standardized data reduces the work load and decreases inconsistency in similarity assessment with different data processing methods. In addition, conformity algorithm does not need weight assignment to each trait, thus can eliminate potential subjective impacts on traits or data and guarantee integrity of information and reliability of evaluation results.
Keywords: Conformity algorithm; Comprehensive evaluation; Computer simulation; Mahalanobis distance
在农学和生物学领域中, 常会遇到对研究对象的个体或群体进行评价的问题。所谓评价就是参照一定的标准, 评判与比较研究对象的价值或优劣的一种认知和决策过程, 如育种过程中对目标品种(系)与参照品种优劣程度的评价、农作物产品品质分级标准的归属或模拟试验中统计数与参数(期望值、标准值)接近程度的评价。依据个体
[1]
(或群体)的多个性状(指标), 以有利于从整体的角度客观、合理、公正地全面评价。所谓多指标综合评价方法, 就是把不同方面的多个指标的信息汇集成一个综合指标, 来反映被评价对象的整体情况。这包括灰色关联度法、TOPSIS法、主成分分析法等[2-6]。多指标综合评价方法一般包括评价指标选择、构建指标体系、选择综合评价模型、
本研究由国家农业信息化工程技术研究中心开放课题“小麦育种材料评价研究”项目资助。 This study was supported by open project of the National Agricultural Information Engineering Center.
*
通讯作者(Corresponding author): 顾世梁, E-mail: slgu@yzu.edu.cn, Tel: 0514-87979358
第一作者联系方式: E-mail: zhanghui881007@126.com, Tel: 18606517137
Received(收稿日期): 2015-02-05; Accepted(接受日期): 2015-09-06; Published online(网络出版日期): 2015-10-13. URL: http://www.cnki.net/kcms/detail/11.1809.S.20151013.1508.004.html