我正在根据观看的平均时间将聚类应用于视频的id变量,该时间是整数类型。这是我的代码:
myclusterdata = data[c(2,11)] #column 2 -->id
gower.dist <- daisy(xx, metric = c("gower"))
class(gower.dist)
#Divisive Clustering (Top-Down)
divisive.clust <- diana(as.matrix(gower.dist), diss = TRUE, keep.diss = TRUE)
plot(divisive.clust, main = "Divisive")
#to which cluster each video belongs
cutree_object = cutree(divisive.clust, 3)
library(cluster)
clusplot(myclusterdata,
cutree_object,
lines = 0,
shade = TRUE,
color = TRUE,
labels = 2,
plotchar = FALSE,
span = TRUE,
main = "Videos clusters based on average time watched",
xlab= "Videos",
ylab= "average time")
尽管我之前已经尝试过该代码并且可以工作,但是现在当我重新运行它时,在运行最后一个绘图命令后,我会显示此错误:
plot.window(...)中的错误:需要有限的“ xlim”值此外: 警告消息:在sqrt(detA * pmax(0,yl2-y ^ 2))中:产生了NaNs
自从上次工作以来,我一直没有更改代码中的内容,那么可能是什么错误? 根据平均观看时间对视频ID进行聚类是否有意义,或者可以对视频数据(例如观看次数和展示次数)进行其他分析工作?