(数据分析)距离判别法-bayes.ppt

5.1距离判别(P159)

本章判别分析discriminateanalysis5.1.1.判别分析训练样本训练样本5.1.2两个总体的距离判别

5.1.3判别准则的评价

例5.1某气象站预报某地区有无春旱的观测资料中，表5.1某地区有无春旱的观测数据解在dataf1;

inputy$x1x2;

cards;

y124.8-2.0

y124.1-2.4

y126.6-3.0

y123.5-1.9

y125.5-2.1

y127.4-3.1y222.1-0.7

y221.6-1.4

y222.0-0.8

y222.8-1.6

y222.7-1.5

y221.5-1.0

y222.1-1.2

y221.4-1.3

;procdiscrimdata=f1 outstat=varstatout=postoutcross=cross method=normal pool=test listcrosslist short;*wcovpcovall;classy;varx1x2;priorsequal;*proportional;run;*p385;语句说明见p385~388.在例5.21991年全国各省、区、市城镇居民月平均消费情况见表5.2.1～20号省份为第一类，记为表5.21991年全国30个省区市城镇居民月平均消费*jiaocaidezuofaruxia;procdiscrimdata=ptestdata=ptestoutstat=aout=b1outcross=b2testout=c method=normal pool=yes listcrosslisttestlistwcov;classy;varx1-x8;priorsequal;*proportional;testclassy;run;procprintdata=a;run;procprintdata=b1;run;procprintdata=b2;run;procprintdata=c;run;解用距离判别法，假定两总体5.1.4多个总体的距离判别

1.总体协方差矩阵相等：2.总体协方差矩阵例5.31991年全国各省、区、市城镇居民月平均收入情况见表5.3.1～11号省份为第一类表5.31991年全国30个省区市城镇居民月平均收入解在datap;

inputy$x1-x9;

cards;

y1170.03110.259.768.384.4926.8016.4411.90.41

y1141.5582.5850.9813.49.3321.3012.369.211.05

y1119.4083.3353.3911.07.5217.3011.7912.00.70y1194.53107.860.2415.68.8831.0021.0111.80.16

y1130.4686.2152.3015.910.520.6112.149.610.47

y1119.2985.4153.0213.18.4413.8716.478.380.51

y1134.4698.6148.188.904.3421.4926.1213.64.56

y1143.7999.9745.606.301.5618.6729.4911.83.82y1128.0574.9650.1313.99.6216.14

10.1814.51.21

y1127.4193.5450.5710.55.8719.41

21.2012.60.90

y1122.96101.469.706.303.8611.30

18.965.624.62

y2102.4971.7247.729.426.9613.12

7.906.660.61

y2106.1476.2746.199.656.279.655

20.106.970.96y210