Fixing the order of a Sankey flow graph in R / networkD3 package

问题

I would like to visualize a rank change (i.e. change of the relative order) of US states using a Sankey flow graph. I'm using the networkd3 package and came up with the following:

library(dplyr)
library(networkD3)

df <- data_frame(origins=state.name[1:10], destinations=state.name[1:10])
lab <- c(df$origins, df$destinations)
nodes <- data.frame(node=c(0:9), name=lab)
links <- data.frame(source=c(0:9), target=c(10:19), value=rep(1,10))
sankeyNetwork(Links = links, Nodes = nodes, Source = 'source', 
              Target = 'target', Value = 'value', NodeID = 'name')

This snippet produces the following graph: Sankey flow graph of US states

I can change the relative order by hand now. However, I wonder whether it is possible to fix the order on the right-hand side and put e.g. Alabama on rank 3, California on rank 1, etc ...

回答1:

If you set iterations = 0 in sankeyNetwork(), you will effectively disable the algorithm which automatically determines the node placement (which is the primary purpose of the sankeyNetwork() function), and the nodes will be placed in the order that they appear in the Nodes dataframe.

library(networkD3)

states <- state.name[1:10]
ranks <- sample.int(10)

nodes <- data.frame(name = c(states, states[ranks]))
links <- data.frame(source = 1:10 - 1, target = order(ranks) + 10 - 1, value = 1)

sankeyNetwork(Links = links, Nodes = nodes, Source = 'source', 
              Target = 'target', Value = 'value', NodeID = 'name',
              iterations = 0)

来源：https://stackoverflow.com/questions/52229334/fixing-the-order-of-a-sankey-flow-graph-in-r-networkd3-package