Search code examples
rdplyrggnetwork

Unique IDs for pairs based on two columns

Hi I am trying to create a directed network based on this data:

ID Order Name
22 1 AA
22 2 BB
22 3 CC
33 1 AA
33 2 GG
44 1 AA
55 1 AA
55 2 BB

The order is the directional order in which they should be connected. So AA ->BB->CC And the width of the edges should be the count of those connections. So far I was able to make a network but the network is also counting and displaying AA->CC as a connection. My hope is that by making a table with unique ID pairs like the one below I will be able to display and count only the correct connections in the network.

ID Name
1 AA
1 BB
2 BB
2 CC
3 AA
3 GG
4 AA
4 AA
5 AA
5 BB

Any advice on how to achieve this would be great!

Solution

  • I thought, what you want to plot your network looks like this. For each ID, we look if there is a flow from AA --> BB or BB --> CC that is unique inside this ID and count them in weight.

    from to weight
    AA BB 2
    AA GG 1
    BB CC 1

    which can be plotted

    out

    Code

    library(dplyr)
library(igraph)

data <- data.frame(ID = c(22,22,22,33,33,44,55,55), Order = c(1,2,3,1,2,1,1,1), Name = c("AA", "BB", "CC", "AA", "GG", "AA", "AA", "BB"))

# edge list with consecutive connections only
edge_weights <- data %>%
  arrange(ID, Order) %>%
  group_by(ID) %>%
  mutate(to = lead(Name)) %>%
  filter(!is.na(to)) %>%
  group_by(from = Name, to) %>%
  summarise(weight = n(), .groups = 'drop')

g <- graph_from_data_frame(edge_weights, directed = TRUE)

plot(g,
     layout = layout_with_fr(g),
     edge.width = E(g)$weight * 2,  # Make edge width proportional to weight
     edge.arrow.size = 0.5,
     vertex.size = 30,
     vertex.color = "lightblue",
     vertex.label.color = "black",
     vertex.label.cex = 1.2,
     main = "Directed Network with Connection Weights")