Short Assignment 1: Reading Network Data

Reading in data

got_marriages <- read_csv("_data/got/got_marriages.csv", show_col_types = FALSE)

IGraph from Edgelist

This edgelist comes with edge attributes Type, Notes and Generation.

#Directed = FALSE because marriages are intrinsically symmetric:
got_marriages.net <- graph_from_data_frame(got_marriages,
                                           directed = FALSE)
got_marriages.net

IGRAPH 98de1a9 UN-- 20 255 -- 
+ attr: name (v/c), Type (e/c), Notes (e/c), Generation (e/c)
+ edges from 98de1a9 (vertex names):
 [1] Targaryen--Stark       Baratheon--Martell     Baratheon--Stark      
 [4] Martell  --Essos       Martell  --Reach       Martell  --Essos      
 [7] Martell  --Essos       Martell  --Septa       Martell  --Dorne      
[10] Targaryen--Martell     Targaryen--Essos       Targaryen--Essos      
[13] Baratheon--Lannister   Baratheon--Vale        Baratheon--Riverlands 
[16] Baratheon--Crownlands  Baratheon--Reach       Baratheon--Westerlands
[19] Baratheon--Crownlands  Lannister--Lannister   Baratheon--Reach      
[22] Baratheon--Tyrell      Tyrell   --Reach       Tyrell   --Reach      
+ ... omitted several edges

These are the vertexes (e.g. nodes). They represent a blend of land areas, administrative zones, and family Houses:

V(got_marriages.net)

+ 20/20 vertices, named, from 98de1a9:
 [1] Targaryen   Baratheon   Martell     Lannister   Tyrell      Reach      
 [7] North       Riverlands  Westerlands Stark       Vale        Arryn      
[13] Tully       Frey        Crownlands  Stormlands  Essos       Septa      
[19] Dorne       Beyond Wall

And these are the types of edges we have:

unique(E(got_marriages.net)$Type)

[1] "Married" "Engaged" "Affair" 

Plot

Here is the most basic plot, before assigning $Type to $weight

plot(got_marriages.net,
            arrow.mode="-")

Here we add each marriage relationship, including affairs, to the weight value which has meaning to the igraph package

E(got_marriages.net)$weight<-E(got_marriages.net)$Type

After weight has been added, the graph looks much different:

plot(got_marriages.net,
            arrow.mode="-")

Warning in (function (graph, coords = NULL, dim = 2, niter = 500, start.temp =
sqrt(vcount(graph)), : NAs introduced by coercion

I want to color the edges according to specific marriage types:

#Define colors for each type of marriage
colors <- c(Married = "red", 
            Engaged = "green", 
            Affair = "blue")

#Match() matches "Married" with "red", and so on
E(got_marriages.net)$color <- colors[match(E(got_marriages.net)$Type, names(colors))]

plot(got_marriages.net,
     arrow.mode="-",
     edge.color=E(got_marriages.net)$color,
     vertex.label.family="Helvetica",
     main="Game of Thrones Marriage Network"
    )

Warning in (function (graph, coords = NULL, dim = 2, niter = 500, start.temp =
sqrt(vcount(graph)), : NAs introduced by coercion

legend("bottomleft", legend=c("Married", "Engaged", "Affair"), fill=colors, title="Type of marriage", cex=0.8)

Here is an alternative, circular layout that in my opinion is not as useful because it does not highlight the social isolation of Beyond Wall:

layout1 <- layout_in_circle(got_marriages.net)

plot(got_marriages.net,
     arrow.mode="-",
     vertex.label.family="Helvetica",
     vertex.label.cex=.5,
     edge.color=E(got_marriages.net)$color,
     edge.width=1,
     main="Game of Thrones Marriage Network",
     layout = layout1
     )
legend("bottomleft", legend=c("Married", "Engaged", "Affair"), fill=colors, title="Type of marriage", cex=0.8)

Statnet

I am assuming “loops” are cases where Lannister – Lannister, or similar. So I’ll be setting that to TRUE. And multiple to TRUE as well.

got_marriages.stat <- as.network(got_marriages, loops = TRUE, multiple = TRUE)
print(got_marriages.stat)

 Network attributes:
  vertices = 20 
  directed = TRUE 
  hyper = FALSE 
  loops = TRUE 
  multiple = TRUE 
  bipartite = FALSE 
  total edges= 255 
    missing edges= 0 
    non-missing edges= 255 

 Vertex attribute names: 
    vertex.names 

 Edge attribute names: 
    Generation Notes Type 

This is a network object rather than a igraph object we were using before.

class(got_marriages.stat)

[1] "network"

class(got_marriages.net)

[1] "igraph"

Plotting is ugly, but functional:

plot(got_marriages.stat,
     displaylabels = T)