我可以使用以下内容生成一个可绘制的树:
library(factoextra)
library(plotly)
hcl <- hclust(
dist(mtcars, method = "euclidean"),
method = "complete")
ggplotly(
fviz_dend(
hcl,
k = 5,
show_labels = FALSE,
type = "phylogenic",
phylo_layout = "layout_as_tree"
)
)
fviz_dend
有一个显示标签的选项,但是当有数百个点时,这变得很难看。相反,我希望能够将鼠标悬停在该点上并查看存储在hcl$labels
中的标签。目前,悬停仅显示坐标和颜色代码:
答案 0 :(得分:1)
我不经常使用plotly软件包,希望有人会提供更好的解决方案,但以下工作(无论如何在我的机器上)都适用于系统树:
k = 5 # change this based on number of groups for cutting the tree
gp <- ggplotly(
fviz_dend(
hcl,
k = k,
show_labels = TRUE, # leave this as the default TRUE, so that the labels
# are passed to the plotly object; we can remove them
# in the next step.
type = "phylogenic",
phylo_layout = "layout_as_tree")
)
# remove hover text for line segments
gp$x$data[[1]]$text <- NA
# for each group, assign the label's text value to the point's text value,
# then remove the label
for(i in seq(k, 1)){
gp$x$data[[i+1]]$text <- gp$x$data[[i+1+k]]$text
gp$x$data[[i + 1 + k]] <- NULL
}
gp
注意:矩形树创建的特定层数与上面显示的有所不同,因此系统树的解决方案将无法直接应用。但是我认为您的用例是系统发育的。