Visualize a social network to show how often a user is mentioned in R [closed]

问题

Given a data frame as following:

v1    v2  v3     v4   v5
tom   A    pinky  A   3
ben   B    hugo   C   2
lily  A    tom    A   1
...

Which denotes that v1 from group v2 has mentioned v3 from group v4 for v5 times. For instance, tom from group A has mentioned pinky from group A for 3 times. Now I'd like to plot a social network, each user denoted by a point and its size is proportional to the times he or she has been mentioned totally. And there is a line linkage two points if they have mentioned each other mutually or unilaterally.

As I look into the ggplot document, I can not find any function to do it.

Do you have any idea? Thanks in advance!

EDIT:

Here is the graph I get so far:

回答1:

library(igraph)
library(dplyr)

# create example dataset
dt = data.frame(v1 = c("tom", "ben", "lilly", "mark"),
                v2 = c("A","B","A","C"),
                v3 = c("pinky", "hugo", "tom", "pinky"),
                v4 = c("A","D","A","A"),
                v5 = c(20,10,15,15),
                stringsAsFactors = F)

dt

#      v1 v2    v3 v4 v5
# 1   tom  A pinky  A 20
# 2   ben  B  hugo  D 10
# 3 lilly  A   tom  A 15
# 4  mark  C pinky  A 15


# select columns of names to use for the graph
dt_graph = dt %>% select(v1,v3)

# create the graph
g = graph.data.frame(dt_graph)

# count number of times names were mentioned
dt_times_mentioned =
  dt %>%
  group_by(v3) %>%
  summarise(times = sum(v5))

dt_times_mentioned

#      v3 times
#   (chr) (dbl)
# 1  hugo    10
# 2 pinky    35
# 3   tom    15


# join back to the vertex names to include names in the graph that were not mentioned
dt_weights =
  data.frame(names = names(V(g)), stringsAsFactors = F) %>%
  left_join(dt_times_mentioned, by=c("names"="v3")) %>%
  mutate(times = ifelse(is.na(times), 0, times))

dt_weights

#   names times
# 1   tom    15
# 2   ben     0
# 3 lilly     0
# 4  mark     0
# 5 pinky    35
# 6  hugo    10


# create two datasets based on 1st and 2nd column of names
dt1 = dt %>% select(names=v1, group=v2) 
dt2 = dt %>% select(names=v3, group=v4)


# get distinct names and their group values
dt_group = 
  dt1 %>% rbind(dt2) %>% distinct() %>% 
  mutate(color = colors()[as.numeric(factor(group))+5]) # get colours from group values

#(note that the +5 above is an arbitrary value for this example in order to get colors easy to distinguish. this is not really needed. if you don't have many groups you can manually set the colors)

dt_group

#   names group         color
# 1   tom     A antiquewhite3
# 2   ben     B antiquewhite4
# 3 lilly     A antiquewhite3
# 4  mark     C    aquamarine
# 5 pinky     A antiquewhite3
# 6  hugo     D   aquamarine1


# plot the graph
plot(g, vertex.size = dt_weights$times, vertex.color = dt_group$color)

# add legend
legend(1.5, 1.5,
       legend=unique(dt_group$group),
       pch=19,
       col=unique(dt_group$color),
       title = "Colors - Groups")

来源：https://stackoverflow.com/questions/34973314/visualize-a-social-network-to-show-how-often-a-user-is-mentioned-in-r

标签

ggplot2

data-visualization

social-networking