Please help me with igraph to pass my course.

Usage
1

I am writing to you because of the greatest need.
I am currently learning to use RStudio, but have more difficulties with it.
But I have to show something in 2 weeks to pass the course.
I have to analyze the “Newcomb Fratnerity” dataset. However, I am already failing to create a network.
Could you show me how I create a network in Rstudio (RMarkdown) from it step by step?
That would be really nice of you.
Or at least tell me whats wrong with this comment: gnewc1 ← graph_from_adjacency_matrix(NEWC0, mode=c(“directed”), weighted=NULL, diag=FALSE, add.colnames=NULL, add.rownames=NA)

With best regards

Taqir

2

Dear Taqir,
could you provide more information, about the data that the objecct NEWC0 contains?
Best
Christian

3

NEWComb
NEWComb716×441 12.4 KB

The dataset look like this.

4

So, is this the dataset from David Schoch’s repo? What did you try so far? What error messages did you get?
When I tried with a minimal example with the first 5 rows/columns from the dataframe you show here, it worked for me:

NEW0 <- matrix(data=c(
  0,9,14,7,11,
  13,0,12,1,8,
  13,9,0,5,4,
  9,1,13,0,7,
  2,9,12,5,0), ncol=5)

gnewc1 <- graph_from_adjacency_matrix(NEW0, mode=c("directed"), weighted=NULL, diag=FALSE, add.colnames=NULL, add.rownames=NA)
2 Likes
5

Yeah it finally work.

First I recoded it to make it look like this:
NEWC0$“1” ← ifelse(NEWC0$“1” <=3 & NEWC0$“1” != 0, 1, 0)
etc.

then I do this:
NEWC0 ← as.matrix(NEWC0)

gnewc0 ← graph_from_adjacency_matrix(NEWC0)
plot(gnewc0)

and it worked.
Now I have to rewrite it so that the values of the dataset indicate the edge strength so that you can see which connections are strongest.

6

Glad that it finally worked although the recoding seems a bit unnecessary to me. To read in the weights you can specify the weighted=TRUE argument. For plotting you could then assign the weights as the edge width and it looks like below.

gnewc1 <- graph_from_adjacency_matrix(NEW0, mode=c("directed"), weighted=TRUE, 
diag=FALSE,  add.colnames=NULL, add.rownames=NA)
E(gnewc1)$width <- E(gnewc1)$weight /2
plot(gnewc1)

Rplot
Rplot1009×524 4.27 KB

1 Like
7

Thanks, that helped!

Now I just have to beautify everything. At the moment it looks more like this:

Netzwerk1
Netzwerk1735×447 10.9 KB

8

Check out this amazing tutorial: Network Analysis and Visualization with R and igraph (kateto.net) by Katherine Ognyanova
At section 5 she describes nicely how to plot networks in igraph. You can play around with edge size, node/label size as well as with the layout.

1 Like