首页 > itarticle > a real network dataset analysis

a real network dataset analysis

admin 11月 14, 2020 0

A real network dataset analysis

1
 
library(igraph)

graph sample

1
2
3
4
5
6
7
8
 
g <- make_graph("Zachary")
plot(g,layout = layout.fruchterman.reingold, 
     vertex.size = 10,
    vertex.color = V(g)$color, 
     #vertex.label.color = NA,
     edge.color = grey(0.5),
    edge.arrow.mode = "-")

png

1
2
3
4
5
6
7
8
9
 
#read gml (dataset : http://www-personal.umich.edu/~mejn/netdata/football.zip)
gg=read_graph(file = "football.gml",format = "gml")
plot(gg,layout = layout.fruchterman.reingold, 
     vertex.size = 10,
    vertex.color = V(gg)$color, 
     #vertex.label.color = NA,
     #
     edge.color = grey(0.5),
    edge.arrow.mode = "-")

png

community

1
2
 
#com = walktrap.community(gg, steps = 5)
com=cluster_walktrap(gg, weights = E(gg)$weight, steps = 4, merges = TRUE, modularity = TRUE, membership = TRUE)
 
1
2
3
4
5
6
7
8
9
10
11
12
 
#subgroup = split(com$merges, com$membership)
## subgroup
V(gg)$sg = com$membership+1
V(gg)$color = rainbow(max(V(gg)$sg))[V(gg)$sg]
## png("net_walktrap.png", width = 500, height = 500)
par(mar = c(0, 0, 0, 0))
set.seed(14)
plot(gg, layout = layout.fruchterman.reingold, vertex.size = 5,
    vertex.color = V(gg)$color, vertex.label = NA, edge.color = grey(0.5),
    edge.arrow.mode = "-")
## dev.off()
print(max(V(gg)$sg)-1)
 
[1] 10

png

subgraph density

1
 
subgraph_1 = split(com$merges, com$membership)
 
Warning message in split.default(com$merges, com$membership):
“data length is not a multiple of split variable”

1
 
subgraph_1[[8]]
 
20 19 4 13 82 62 141 162 207 221 35 96 27 128 143 173 172 187 223 226

1
 
graph.density(gg)

0.0935163996948894

1
2
 
sg8=subgraph(gg,V(gg)[sg==8])
graph.density(sg8)
 
Warning message in .Call("R_igraph_subgraph", graph, as.igraph.vs(graph, v) - 1, :
“At structural_properties.c:1945 :igraph_subgraph is deprecated from igraph 0.6, use igraph_induced_subgraph instead”

0.727272727272727

1
2
3
4
5
6
7
8
 
library(ggplot2)
sg_d=c()
for(i in (1:10)){
    #sg from 2
    sg_t=subgraph(gg,V(gg)[sg==(i+1)])
    d=graph.density(sg_t)
    sg_d[i]=d
}
 
1
2
3
4
5
6
 
data_d=data.frame(com=c(as.character(1:10)),density=sg_d)
data_d$com = factor(data_d$com, levels=data_d$com)
p = ggplot(data_d, aes(x=com, y=density, fill=com,group=factor(1))) +geom_bar(stat="identity", width=0.5, position=position_dodge(0.6)) + #值越大越离得远
    geom_text(aes(label = com, vjust = -0.8, hjust = 0.5, color = com), show.legend = FALSE)
#p=ggplot(data_d, aes(sg_d)) + geom_bar()
plot(p)

png

node betweenness

1
2
3
4
5
6
7
8
9
10
 
V(gg)$bte = betweenness(gg, directed = F)
## png("net_betweenness.png", width = 500, height = 500)
par(mar = c(0, 2, 0, 0))
#plot(V(gg)$bte,xlab="index",ylab="betweenness",col="green")
## dev.off()
data_v=data.frame(index=1:length(V(gg)),betweenness=V(gg)$bte)
p = ggplot(data_v,aes(x=index,y=betweenness,colour="index")) +geom_point(size=3) + #
    geom_text(aes(label = index,vjust = -0.8, hjust = 0.5), show.legend = FALSE)
#p=ggplot(data_d, aes(sg_d)) + geom_bar()
plot(p)

png

1
2
3
4
5
6
7
 
V(gg)$size = 5
V(gg)[bte>=150]$size = 15
#V(gg)$label=NA
#V(gg)[bte>=150]$label=V(gg)[bte>=150]$name
#V(gg)[bte>=150]$label
V(gg)$cex=1
V(gg)[bte>=150]$cex=2
 
1
2
3
4
5
6
7
8
9
 
## png("net_walktrap_betweenness.png", width = 500, height = 500)
par(mar = c(0, 0, 0, 0))
set.seed(14)
plot(gg, layout = layout.fruchterman.reingold, vertex.size = V(gg)$size,
    vertex.color = V(gg)$color, vertex.label = V(gg)$label,
    vertex.label.cex=V(gg)$cex, edge.color = grey(0.5),
    edge.arrow.mode = "-")
## dev.off()
V(gg)[bte>=150]$label
 
'BrighamYoung' 'CentralMichigan' 'LouisianaTech' 'NewMexicoState' 'Navy' 'NotreDame' 'Cincinnati'

png

a sample of recommendation

1
2
 
neighbors(gg,V(gg)$label=="BrighamYoung")
V(gg)[5]$label
 
+ 12/115 vertices:
 [1]   2   5  10  17  24  34  36  42  66  91  94 105


'NewMexico'

1
2
3
 
#hashmap instead
#map=c()
#names(map)=c()

based on betweenness

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
 
#neighbors(gg,V(gg)$label=="BrighamYoung",mode = c("out", "in", "all", "total"))
#tmp = neighbors(gg,V(gg)$label=="BrighamYoung") %in% neighbors(gg,V(gg)$label=="NewMexico")
#tmp
#tmp1 = setdiff(neighbors(gg,V(gg)$label=="BrighamYoung") , neighbors(gg,V(gg)$label=="NewMexico"))
tmp1 = setdiff(V(gg),neighbors(gg,V(gg)$label=="NewMexico"))
tmp1_bte=V(gg)[tmp1]$bte
#t=table(tmp1_bte,names=tmp1)
#t
t=tmp1_bte
names(t)=tmp1
listall = sort(t, dec = T)
top = listall[1:length(t)]
#top
#names(which(t[1,]==1))
t1 = top[1:5]
t1_names=V(gg)[as.integer(names(t1))]$label
names(t1)=t1_names
#t1[as.character(top[1])]
t1
 
NotreDame 215.98577427622
Navy 187.826344074865
LouisianaTech 185.647805653186
CentralMichigan 162.228191494366
Cincinnati 153.529962162747

based on common friends

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
 
# NewMexico好友
t_n=neighbors(gg,V(gg)$label=="NewMexico")
# 非好友
t_nn=setdiff(V(gg),t_n)
t_nn=t_nn[-which(t_nn==5)]
# 共同好友计算
t_nn_ld=c()
t_nn_=c()
i=0
for(t_ in t_nn){
    d=sum(neighbors(gg,t_) %in% t_n)
    if(d>=1){
        i=i+1
        t_nn_ld[i]=d
        t_nn_[i]=t_
    }else{
        
    }
}
t_nn_linkdegree_dic=t_nn_ld
names(t_nn_linkdegree_dic)=t_nn_
 
1
2
3
4
5
 
t_nn_ld_=sort(t_nn_linkdegree_dic,dec=T)
if(length(t_nn_ld_)>=5){
    t_nn_ld_=t_nn_ld_[1:5]
}
#t_nn_ld_
 
1
2
 
print("rec :")
t_nn_ld_
 
[1] "rec :"


91 5
12 4
23 3
68 3
79 3

1
2
3
4
5
 
print("names :")
for (i in (1:length(t_nn_ld_))){
    t_nn_ld_[i]=V(gg)[as.integer(names(t_nn_ld_[i]))]$label
}
t_nn_ld_
 
[1] "names :"


91 'UtahState'
12 'NorthTexas'
23 'Arizona'
68 'Nevada'
79 'WashingtonState'

Reference

http://cos.name/2011/04/exploring-renren-social-network/

微海报分享

近期文章

近期评论

标签

热门

文章归档

分类目录

功能