Visualize a Large Graph Network in R

Question

I have around 100,000 rows in a dataframe and I need to visualize it using a network graph in R. However, since there is too much data, it is very difficult to analyze visually and I am not sure how to do this since I am new in R.

This is what I am aiming for:

And this is what my df looks like:

Location	Manager
L1	M1
L2	M3
L76	M1
L34	M1
L45	M1
L18	M4
L98	M7
L145	M4
L134	M1
L22	M5
L5	M7
L56	M7
L11	M8
L76	M5

For example, location L22 should be connected to location L76 since they have M5 in common, and so on. I also want the weight of the line connecting these locations to be based on the number of managers they have in common.

Thanks!

But surely a graph of the size you describe would be uninterpretable visually? What is the message that you want to convey with your visualisation? — Allan Cameron, Aug 29 '22 at 21:19
A place to start could br creating a weighted adjaceny matrix with `adj = tcrossprod(table(d))` . Then just use standard graph tools e.g. `igraph` — user20650, Aug 29 '22 at 23:32
... and *lots* of examples for changing node size and edge weight : you'll find a bunch of examples at https://stackoverflow.com/search?q=%5Br%5D+igraph+edge+thickness and https://stackoverflow.com/search?q=%5Br%5D+igraph+node+size — user20650, Aug 29 '22 at 23:37
Since your example desired graph came from [Previous Answer](https://stackoverflow.com/a/64729226/4752675), did you try the code there? What did not work for you? — G5W, Oct 03 '22 at 17:02

score 0 · Accepted Answer · answered Aug 30 '22 at 09:27

I guess you can use the igraph package like below

library(igraph)
g <- simplify(
  graph_from_data_frame(
    do. Call(
      rbind,
      lapply(
        split(df, ~Manager),
        function(v) {
          with(
            v,
            if (length(Location) > 1) {
              make_full_graph(length(Location)) %>%
                set_vertex_attr(name = "name", value = Location) %>%
                set_edge_attr(name = "width", value = length(Manager)) %>%
                get.data.frame()
            } else {
              data. Frame(from = Location, to = Location, width = 1)
            }
          )
        }
      )
    ),
    directed = FALSE
  ),
  edge.attr.comb = "sum"
)

and you will obtain

> g
IGRAPH d49cf33 UN-- 13 15 -- 
+ attr: name (v/c), width (e/n)
+ edges from d49cf33 (vertex names):
 [1] L1 --L76  L1 --L34  L1 --L45  L1 --L134 L76--L34  L76--L45  L76--L22
 [8] L76--L134 L34--L45  L34--L134 L45--L134 L18--L145 L98--L5   L98--L56
[15] L5 --L56

and the network plot (run plot(g))

it doesn't work for me, the connections between the locations do not show up at all — melisa, Aug 31 '22 at 14:30
@melisa you'd better show how your desired output looks like. I have no clue what you are after. — ThomasIsCoding, Aug 31 '22 at 14:40

Visualize a Large Graph Network in R

1 Answers1