《2016年北京工业大学数学建模竞赛初赛试题B题 [在线提供]》.doc

2007年北京工业大学数学建模竞赛初赛试题B题： 化验结果的处理 题解 Mahalanobis距离判别。根据两个母体样本计算出他们的均值向量和协方差阵，求取待测样本x对两个样本的Mahalanobis距离，二者取差值，判断离那个母体近似。 设x,y是从均值为μ，协方差阵为Σ的总体A中抽取的样本，则总体A内两点x与y的Mahalanobis距离定义为.定义样本x与A的Mahalanobis距离为。在现实中，母体的均值向量和协方差阵由样本的均值向量和协方差阵来代替：设，，……是来自母体A的个样本，，，……是来自母体A的个样本，则样本的均值与协方差为，，对于待测样本x，如果两个母体样本的协方差相同，由得到判别函数为，其中，其判别准则是 。如果两个母体样本协方差不同，即，对于样本x判别函数定义为：，。 其次考虑用另外一种方法求取解决办法——Fisher判别法，即按类内方差尽量小，类间方差尽量大的准则来求判断函数。 设两个总体A、B的均值和协方差阵分别是、和、，对任一测样本x，设它的判别函数为，并假设,,使满足类内偏差平方和最小，而类间偏差平方和最大，其中。即要满足最大，若，则，否则。通过推导得出判别函数，其中，，当，,否则。 6．模型求解 利用模型求解时通过R软件将以上两种算法编写成程序代码，通过手动输入样本，利用计算机进行求解，程序清单如下： Mahalanobis距离判别： A-matrix(c(166,15.8,24.5,700,112,179,513,185,15.7,31.5,701,125,184,427,193,9.80,25.9,541,163,128,642,159,14.2,39.7,896,99.2,239,726,226,16.2,23.8,606,152,70.3,218,171,9.29,9.29,307,187,45.5,257,201,13.3,26.6,551,101,49.4,141,147,14.5,30.0,659,102,154,680,172,8.85,7.86,551,75.7,98.4,318,156,11.5,32.5,639,107,103,552,132,15.9,17.7,578,92.4,1314,1372,182,11.3,11.3,767,111,264,672,186,9.26,37.1,958,233,73.0,347,162,8.23,27.1,625,108,62.4,465,150,6.63,21.0,627,140,179,639,159,10.7,11.7,612,190,98.5,390,117,16.1,7.04,988,95.5,136,572,181,10.1,4.04,1437,184,101,542,146,20.7,23.8,1232,128,150,1092,42.3,10.3,9.70,629,93.7,439,888,28.2,12.4,53.1,370,44.1,454,852,154,13.8,53.3,621,105,160,723,179,12.2,17.9,1139,150,45.2,218,13.5,3.36,16.8,135,32.6,51.6,182,175,5.84,24.9,807,123,55.6,126,113,15.8,47.3,626,53.6,168,627,50.5,11.6,6.30,608,58.9,58.9,139,78.6,14.6,9.70,421,70.8,133,464,90.0,3.27,8.17,622,52.3,770,852,178,28.8,32.4,992,112,70.2,169),ncol=7, byrow=T) B-matrix(c(213,19.1,36.2,2220,249,40.0,168,170,13.9,29.8,1285,226,47.9,330,162,13.2,19.8,1521,166,36.2,133,203,13.0,90.8,1544,162,98.90,394,167,13.1,14.1,2278,212,46.3,134,164,12.9,18.6,2993,197,36.3,94.5,167,15.0,27.0,2056,260,64.6,237,158,14.4,37.0,1025,101,44.6,72.5,133,22.8,31.0,1633,401,180,899,156,135,322,6747,1090,228,810,169,8.00,308,1068,99.1,53.0,289,247,17.3,8.65,2554,241,77.9,373,166,8.10,62.8,1233,252,134,649,209,6.43,86.9,2