logistic回归的问题

shusun

在用R做logistic回归时出现问题，出现

1: glm.fit:算法没有聚合

2: glm.fit:拟合機率算出来是数值零或一的警告信息，而且回归系数都不显著

程序如下

其中A1，A2，A3，A4是名义数据的处理

x<-rep(c(25,26,27,27,29,30,25,26,27,27,29,30),c(68,55,66,32,21,25, 42,38,20,10,7,4))

A1<-rep(c(1,0,0,0,-1),c(66,95,103,101,23))

A2<-rep(c(0,1,0,0,-1),c(66,95,103,101,23))

A3<-rep(c(0,0,1,0,-1),c(66,95,103,101,23))

A4<-rep(c(0,0,0,1,-1),c(66,95,103,101,23))

y<-rep(c(1,0),c(267,121))

weight<-data.frame(x,A1,A2,A3,A4,y)

lm2<-glm(y~x+A1+A2+A3+A4,family=binomial,data=weight)

summary(lm2)

求指教。。。谢谢

nan.xiao

http://cos.name/cn/topic/105759

http://cos.name/cn/topic/109544