4

我试图使用gofpackage 中的函数进行样本外预测btergm。从测试集中计算一条precision-recall曲线的AUC值时,我得到的结果是1.012909,这在理论上似乎是不可能的。

我该如何解释结果,或者我做错了什么。谢谢!这是我的代码:

network <- readRDS(url("https://www.dropbox.com/s/zxhqxa8h9awkzpd/network.rds?dl=1"))
model.3<- btergm(network[1:9]~edges+gwodegree(1, fixed=TRUE)+transitiveties + ctriple +
                gwidegree(1, fixed=TRUE)+mutual+gwesp(1.5, fixed=TRUE)+ttriple+
                memory(type="stability")+delrecip,R=1000)
gof.3 <-gof(model.3, nsim = 1000,target=network[[10]],formula = network[9:10]~edges+gwodegree(1,fixed=TRUE) + 
               transitiveties + ctriple +gwidegree(1,fixed=TRUE)+mutual+gwesp(1.5, fixed=TRUE)+ttriple+
               memory(type="stability")+delrecip,coef = coef(model.3), 
               statistics = rocpr)
gof.3[[1]]$auc.pr
4

0 回答 0