Part 1: The Data
source("./Import Scripts/Game of Thrones Interactions.R")
Note: I believe that I have fixed the issue with statnet by commenting out the line “delete.edge.attribute(network_statnet,”na“)” that was after the igraph to statnet conversion. This means that the network now has edges, but I am unsure if they are the correct ones.
This is an edgelist dataset. It has 298 vertices and 9131 edges, it is not directed or bipartite, and it is weighted. The vertices are the people, with attributes regarding the time and location of their appearances, as well as the point of view the story is in for these appearances.
The dataset is the Game of Thrones Interactions dataset, with characters as nodes, and instances of characters being mentioned in the same vicinity are the edges.
Here is the best network plot I have managed to create with this data so far:
V(network_igraph)$size<-5
plot(network_igraph, layout_with_lgl(network_igraph), vertex.label=NA)
Part 2
Betweenness
centr_betw(network_igraph,directed=F)$centralization #centralization
[1] 0.04738167between<-data.frame(igraph::betweenness(network_igraph, directed=FALSE, weights=NA)) #centrality list
head(arrange(between, desc(between)), 10) #top 10 most betweenness central nodes
igraph..betweenness.network_igraph..directed...FALSE..weights...NA.
Robert Baratheon 2201.4513
the Others 1815.0708
Robb Stark 1787.4981
Eddard Stark 1665.9894
Tywin Lannister 1654.3540
Stannis Baratheon 1473.8467
Arya Stark 1458.7710
Jaime Lannister 1179.1043
Jon Snow 935.7536
Roose Bolton 930.4197This list of nodes is fairly consistent with the other lists from other methods of finding centrality, so it is not very surprising- except for maybe Roose Bolton making the list, when he has not appeared in any previous lists of top ten most central.
The betweenness value for the whole network is ~0.05, meaning that these centralities are not concentrated on one or few nodes.
Brokerage
This network is an undirected network, so for this last part I will use the Game of Thrones Like/Dislike network, which is directed and similar to the interactions network.
source("C:/Users/zabea/OneDrive/Documents/DACSS/spring2022/Political Networks/Import Scripts/Game of Thrones LikeDislike.R")
#################################################################################################
"Game of Thrones Like-Dislike.R" has imported the Like-Dislike Network from the Game of Thrones
Dataset.
It is a small, weighted, directed network showing characters affinity for one another. The degree
to which a character likes another is held by the "weight" attribute and ranges between -5 and 5.
The import script has created four objects that represent the network:
-network_adjacency2 (an adjacency matrix)
-network_nodes2 (a dataframe of node attributes)
-network_igraph2 (an igraph object)
-network_statnet2 (a network object compatable with statnet packages like sna & ergm)
Each object name starts, quite generically, with "network_" and ends with the type of object it
is. Note that the names are generic so that this import script is compatable with other scripts
we will use with this course.
################################################################################################The basic data for this network:
#network size, weighted, directed, bipartite
print(network_statnet2)
Network attributes:
vertices = 46
directed = TRUE
hyper = FALSE
loops = FALSE
multiple = FALSE
bipartite = FALSE
total edges= 1143
missing edges= 0
non-missing edges= 1143
Vertex attribute names:
Current.house Former.house vertex.names
Edge attribute names not shown #attributes
#vertices
network::list.vertex.attributes(network_statnet2)
[1] "Current.house" "Former.house" "na" "vertex.names" head(network_statnet2 %v% "vertex.names")
[1] "Lysa.Arryn" "Petyr.Baelish" "Joffrey.Baratheon"
[4] "Margaery.Tyrell" "Renly.Baratheon" "Robert.Baratheon" #edges
network::list.edge.attributes(network_statnet2)
[1] "na" "weight"head(network_statnet2 %e% "weight")
[1] 3 2 1 -1 -1 2There are much less nodes in this network- only 46 compared to our previous 298.
broker<-data.frame(brokerage(network_statnet2, cl =
network_nodes2$Current.house)$z.nli)
network_nodes2<-network_nodes2 %>%
mutate(broker.tot = broker$t,
broker.coord = broker$w_I,
broker.itin = broker$w_O,
broker.rep = broker$b_IO,
broker.gate = broker$b_OI,
broker.lia = broker$b_O)
head(arrange(network_nodes2, desc(network_nodes2$broker.tot)), 10)
Name Current.house Former.house broker.tot
1 Joffrey Baratheon Lannister Baratheon 4.4280731
2 Cersei Lannister Lannister Baratheon 4.3525892
3 Catelyn Stark Stark Tully 3.8996860
4 Robb Stark Stark 3.2580731
5 Daenerys Targaryen Targaryen 2.5598472
6 Theon Greyjoy Greyjoy 1.8427504
7 Tywin Lannister Lannister 1.4464601
8 Sansa Stark Stark 1.0690407
9 Arya Stark Stark 0.9180729
10 Bran Stark Stark 0.9180729
broker.coord broker.itin broker.rep broker.gate broker.lia
1 1.1793625 -0.5165196 3.5503530 3.7333268 3.594505
2 -0.4021732 0.2883724 3.1234143 2.3915193 4.490194
3 -1.1969890 -0.8095261 3.3445779 3.3445779 3.508737
4 -1.1969890 -0.8095261 2.9008395 2.8374483 2.921579
5 -0.5623731 0.5010289 1.2621034 1.2621034 2.623248
6 NaN -1.6744801 0.1706807 0.1706807 2.562505
7 -0.4021732 -1.0531144 1.2326856 1.1716944 1.515227
8 -1.1969890 -2.2856053 1.3794505 1.3160593 1.191010
9 -1.1969890 -2.5316185 1.1258857 1.2526681 1.160107
10 -1.1969890 -2.5316185 1.1258857 1.2526681 1.160107I decided to use Current.house because most people have a current house, while few have a former house.
This, too, is fairly consistant with the centrality lists from the interactions data. Joffrey Baratheon is on top of this list- and was almost universally hated by every other character in the books, save his own mother, who is number two on this list, is Queen for all or almost all of the series, and was also fairly unpopular amongst the other characters. the majority of the people on this list generally have a lot of supporters, or a lot of detractors. It should be noted that five of the ten people on this list are currently Starks, which is also unsurprising.
The books written mostly from the point of view of the Starks, their allies, their enemies, and their enemies turned allies. Three of the remaining five are Lannisters, and since this story has a lot of Stark/Lannister conflicts, tentative allyships, and betrayals it is no surprise that there is a lot of brokerage happening with these families.