I'm new to R and want to create a network diagram basically does this
| postid | userid |
|---|---|
| 3 | 2 |
| 3 | 4 |
| 3 | 5 |
| 1 | 1 |
| 1 | 2 |
| 2 | 1 |
| 2 | 4 |
| -------- | -------------- |
where user1 and user2 are connected to each other and user 2,4,5 is connected to each other too,amd there should be an edge between user 1 and user 4 since they both posted in post 2 basically all i want is to generate a network diagram to show users who talked in the same post.
Thanks in advance
df <- read.table(text="postid userid
3 2
3 4
3 5
1 1
1 2
2 1
2 4", header = TRUE)
library(tidyverse)
library(igraph)
df %>%
left_join( df, by = "postid" ) %>%
select(-postid) %>%
graph_from_data_frame(directed = FALSE) %>%
simplify( remove.multiple = TRUE, remove.loops = TRUE) %>%
plot()
You can try the code below
glst <- lapply(
with(df, split(as.character(userid), postid)),
function(x) graph_from_data_frame(t(combn(x, 2)), directed = FALSE)
)
or
lapply(
with(df, split(as.character(userid), postid)),
function(x) simplify(graph_from_data_frame(expand.grid(x, x), directed = FALSE))
)
which produces a list of graph ojbects
$`1`
IGRAPH 125b424 UN-- 2 1 --
+ attr: name (v/c)
+ edge from 125b424 (vertex names):
[1] 1--2
$`2`
IGRAPH 125bde9 UN-- 2 1 --
+ attr: name (v/c)
+ edge from 125bde9 (vertex names):
[1] 1--4
$`3`
IGRAPH 125bde9 UN-- 3 3 --
+ attr: name (v/c)
+ edges from 125bde9 (vertex names):
[1] 2--4 2--5 4--5
If you want to aggregate all sub-graphs, you can use graph.union
> (g <- do.call(graph.union, glst))
IGRAPH 7c8e074 UN-- 4 5 --
+ attr: name (v/c)
+ edges from 7c8e074 (vertex names):
[1] 4--5 2--5 2--4 1--4 1--2
Data
> dput(df)
structure(list(postid = c(3L, 3L, 3L, 1L, 1L, 2L, 2L), userid = c(2L,
4L, 5L, 1L, 2L, 1L, 4L)), class = "data.frame", row.names = c(NA,
-7L))