I am using rstudio version 0.99.879 and the package igraph version 1.0.1. My question is closely related to this one on the same problem: igraph package in RStudio: Bipartite graph projection error
However, my question is more related to the data structure. Here is a link to an example of the csv.file that I use: https://workupload.com/file/6qhyZqc and the following code:
# Start
set.seed(7)
setwd("C:/Users/Stefan/Desktop/")
data <- read.csv("example.csv", sep=";")
summary(data)
library(igraph)
## using subset function to select 2 variables
data_new <- subset(data, select=c(justification, claimant_function))
data_new
g <- graph.data.frame(data_new, directed = FALSE)
g
col <- c("steelblue1", "white")
shape <- c("circle", "square")
# creating bipartite network
V(g)$type <- FALSE
V(g)$type[V(g)$name %in% data_new[, 1]] <- TRUE
is.bipartite(g)
# TRUE
plot(g, layout = layout.kamada.kawai,
vertex.shape = shape[as.numeric(V(g)$type) + 1],
vertex.color = col[as.numeric(V(g)$type) + 1],
edge.color = 'gray')
# bipartiate projection
one_mode_networks <- bipartite.projection(g)
# Error in .Call("R_igraph_bipartite_projection", graph, types,
# as.integer(probe1), :
# At bipartite.c:198 : Non-bipartite edge found in bipartite projection,
# Invalid value
Everything works out except the projection command. So,the code is not the problem. Perhaps the mistake/problem might be in the data itself. Since I am working with the data already quite a while, I think, I am professionally blinkered. It would be great if someone else could have a look at the posted example data and could give suggestions what the issue might be.
Any help is highly welcomed!
My hunch is that you have a tie between two nodes of the same affiliation. A bit of searching and indeed this is the case.
Notice that you have the string 'other' in both columns of data_new:
data_new[which(data_new[,1] %in% data_new[,2]),1]
[1] "other"
igraph reads the string as the one node in the network. V(g)$type[V(g)$name %in% data_new[, 1]] <- TRUE gives it a value of TRUE for its type.
We can see which tie connects two vertices whose type == 'TRUE':
> i <- which(V(g)$type[match(ends(g,1:ecount(g))[,1],V(g)$name)] == V(g)$type[match(ends(g,1:ecount(g))[,2],V(g)$name)])
> ends(g, i)
[,1] [,2]
[1,] "financial solidity" "other"
The 12th edge, both vertices have type==TRUE.
Simply revalue the strings so that they are not equal and everything runs smoothly.
data_new <- subset(data, select=c(justification, claimant_function))
data_new[which(data_new[,1]=="other"),1] <- "other just"
data_new[which(data_new[,2]=="other"),2] <- "other claim"
g <- graph.data.frame(data_new, directed = FALSE)
# creating bipartite network
V(g)$type <- FALSE
V(g)$type[V(g)$name %in% data_new[, 1]] <- TRUE
is.bipartite(g)
one_mode_networks <- bipartite_projection(g)
Check:
> one_mode_networks
$proj1
IGRAPH UNW- 16 72 --
+ attr: name (v/c), weight (e/n)
+ edges (vertex names):
[1] business --expert/scientist business --public figure
[3] business --media/journalist business --citizen
[5] business --legislative business --ECB
[7] government --media/journalist government --expert/scientist
[9] government --other claim government --legislative
[11] government --ECB government --European Commission
[13] government --other politician/party government --European Parliament
[15] government --citizen government --European Council
+ ... omitted several edges
$proj2
IGRAPH UNW- 16 83 --
+ attr: name (v/c), weight (e/n)
+ edges (vertex names):
[1] political solidarity--monetary solidarity political solidarity--financial solidity
[3] political solidarity--no justification political solidarity--cultural solidarity
[5] political solidarity--sovereignty political solidarity--self interest
[7] political solidarity--economic solidarity political solidarity--human solidarity
[9] financial solidity --social solidarity financial solidity --misuse of solidarity
[11] financial solidity --economic solidarity financial solidity --cultural solidarity
[13] financial solidity --self interest financial solidity --legal regulations
[15] financial solidity --necessity financial solidity --conditionally
+ ... omitted several edges