Get automatically nodes and links from table for sankey diagram

问题

For plotting a sankey diagram nodes and links are required. To get the nodes and links from a data frame one could use for example a count function from the package plyr and use it for each node to count the links between the neighbours but is there another elegant way?

example target, aim is to get nodes and links:

param1 | param2 | param3 |
a      | b      | d      |
w      | c      | d      |
a      | b      | d      |
z      | c      | e      |

#nodes:
nodes = data.frame("name" = 
c(
a, #node 0
w, #node 1
z, #node 2
b, #node 3
c, #node 4
d, #node 5
e  #node 6
))

#links
links = as.data.frame(matrix(c(
0, 3, 2, # from node 0,  to node 3, freq
1, 4, 1,
2, 4, 1,
3, 5, 2,
4, 5, 1,
4, 6, 1,
),
byrow = TRUE, ncol = 3))

回答1:

Using igraph package:

library(dplyr)
library(igraph)

# example data
df1 <- read.table(text="
                  param1 param2 param3
                  a b d
                  w c d
                  a b d
                  z c e", header = TRUE, stringsAsFactors = FALSE)

# make graph
g <- graph_from_data_frame(
  rbind(
    setNames(df1[, 1:2], c("from", "to")),
    setNames(df1[, 2:3], c("from", "to"))))


nodes <- data.frame(id = as.numeric(V(g)),
                    name = V(g)$name)
nodes
#   id name
# 1  1    a
# 2  2    w
# 3  3    z
# 4  4    b
# 5  5    c
# 6  6    d
# 7  7    e

links <- as.data.frame(get.edges(g, E(g))) %>%
  group_by(V1, V2) %>%
  summarise(freq = n()) %>% 
  data.frame()

links
#   V1 V2 freq
# 1  1  4    2
# 2  2  5    1
# 3  3  5    1
# 4  4  6    2
# 5  5  6    1
# 6  5  7    1

来源：https://stackoverflow.com/questions/50600070/get-automatically-nodes-and-links-from-table-for-sankey-diagram