从Flexmix中提取最佳参数：单变量数据

Question

我正在尝试复制书籍HMM中的一个表格：

具体来说，k = 2模型。

我正在使用代码：

dat<-as.data.frame(c(13,  8,  23,   22, 18, 15, 14, 11, 24, 18, 14, 16, 8,  14, 27, 15, 10, 13, 10, 23, 41, 20, 15, 15, 16, 18, 31, 15, 8,  16, 26, 17, 27, 22, 15, 11, 32, 19, 35, 19, 6,  11, 27, 20, 26, 16, 11, 18, 22, 28, 30, 8,  32, 19, 36, 27, 7,  36, 13, 39, 29, 18, 24, 26, 21, 23, 16, 22, 13, 17, 20, 13, 23, 14, 22, 16, 12, 22, 22, 17, 21, 13, 18, 24, 19, 21, 20, 25, 21, 15, 25, 15, 21, 22, 34, 16, 16, 21, 26, 10, 18, 12, 14, 21, 15, 15, 18))
colnames(dat)<-"x"

x2 <- flexmix(x~1, data=dat, k=2,model=FLXMRglm(formula=x~1,family="poisson"))
summary(x2)

检查先验和日志似乎语法正在做我想要的:)。我的问题是如何提取组件意味着lambda（上面的15.777和26.840）？我不认为它们只是每个集群中数据的平均值。

Answer 1

exp(parameters(x2))

您可以使用parameters提取系数，但模型默认使用日志链接功能，因此您需要使用exp转换回原始比例。