桑基图可以用来表示各个节点之间转换
在R中可以直接定义点点之间的关系后使用相应的package画桑基图：
方法一：
- library(riverplot)
- # 构造连接节点的数据框
- edges = data.frame(N1 = paste0(rep(LETTERS[1:4], each = 4), rep(1:5, each = 16)),
- N2 = paste0(rep(LETTERS[1:4], 4), rep(2:6, each = 16)),
- Value = runif(80, min = 2, max = 5) * rep(c(1, 0.8, 0.6, 0.4, 0.3), each = 16),
- stringsAsFactors = F)
- # 筛选80%的记录，以免每个点都对应到4个点
- edges = edges[sample(c(TRUE, FALSE), nrow(edges), replace = TRUE, prob = c(0.8, 0.2)),]
- head(edges)
- nodes = data.frame(ID = unique(c(edges$N1, edges$N2)), stringsAsFactors = FALSE)
- #
- nodes$x = as.integer(substr(nodes$ID, 2, 2))
- nodes$y = as.integer(sapply(substr(nodes$ID, 1, 1), charToRaw)) - 65
- #
- rownames(nodes) = nodes$ID
- head(nodes)
- # 添加颜色
- library(RColorBrewer)
- # 后面加调淡颜色
- palette = paste0(brewer.pal(4, "Set1"), "60")
- # 对每个节点生成相应的格式
- styles = lapply(nodes$y, function(n) {
- list(col = palette[n+1], lty = 0, textcol = "black")
- })
- names(styles) = nodes$ID
- # 以list的结构保存一遍调用
- rp <- list(nodes = nodes, edges = edges, styles = styles)
- class(rp) <- c(class(rp), "riverplot")
- plot(rp, plot_area = 0.95, yscale=0.06)
效果图见：
方法二：
使用包d3Network或者circlize，d3Network是调用D3的画图功能来实现，使用一下代码实现：
- library(d3Network)
- d3links <- edges
- d3nodes <- data.frame(name = unique(c(edges$N1, edges$N2)), stringsAsFactors = FALSE)
- d3nodes$seq <- 0:(nrow(d3nodes) - 1)
- d3links <- merge(d3links, d3nodes, by.x="N1", by.y="name")
- names(d3links)[4] <- "source"
- d3links <- merge(d3links, d3nodes, by.x="N2", by.y="name")
- names(d3links)[5] <- "target"
- names(d3links)[3] <- "value"
- d3links <- subset(d3links, select=c("source", "target", "value"))
- d3nodes <- subset(d3nodes, select=c("name"))
- # 画图并保存为html文件
- d3Sankey(Links = d3links, Nodes = d3nodes, Source = "source",
- Target = "target", Value = "value", NodeID = "name",
- fontsize = 12, nodeWidth = 30, file = "TestSankey.html")
效果图：
备注：
(1)d3Network保存为HTML文件，有交互式效果；riverplot可选择保存为不同的图片类型
(2)操作中感觉使用riverplot更易调整