library(igraph)
setwd("~/Documents/Violence Cascades/")

# plot cascade sizes
data = read.csv('Results/components_dist-91515.csv',header=F)
data = data[order(data$V1),]
sizes = data$V1
counts = data$V2
plot(sizes,counts,log='xy',type='o',lwd=3,
     xlab='Size of Cascade', ylab='Number of Cascades', main='Distribution of Cascade Sizes')


# plot cascades
edges = read.csv('Results/edges-91515.csv',header=F,
                 col.names=c('v1','t1','v2','t2','dist'))
for(id in unique(union(edges$v1,edges$v2))){
  e = edges[union(match(id,edges$v1), match(id,edges$v2)),]
  times = sort(union(e$t1[e$v1==id],e$t2[e$v2==id]))
  if (length(times)>1){
    for(time in times[-1]){
      idx = which(time==times)
      edges$v1[as.numeric(rownames(e))[e$t1==time]] = e$v1[e$t1==time] + (idx-1)/length(times)
      edges$v2[as.numeric(rownames(e))[e$t2==time]] = e$v2[e$t2==time] + (idx-1)/length(times)
    }
  }
}
dag = graph.data.frame(edges[,c(1,3)], directed=TRUE)
table(clusters(dag)$csize)




clusters = clusters(dag)
membership = clusters$membership
csize = clusters$csize
order = rev(order(csize))


i = 4
V = which(clusters(dag)$membership==order[i]) # get all nodes in cluster
cc = induced.subgraph(dag,V)
Vi = vic_ids[V]
Ei = intersect(which(dag_dat_vics$from[realized] %in% Vi),which(dag_dat_vics$to[realized] %in% Vi))
cc_dat = (dag_dat_vics[realized,])[Ei,]

### plot cascade ###
cols = rep('lightblue',vcount(cc))
seed = which(degree(cc,mode='in')==0)
cols[seed] = 'red'
plot(cc,vertex.size=10,edge.arrow.size=0.5,vertex.color=cols,vertex.label.cex=1,
     edge.width=E(cc)$weight*20/max(E(cc)$weight),layout=layout.reingold.tilford(cc,root=seed),
     vertex.label=V(hyp_lcc)$vic.day[Vi])
plot(cc,vertex.size=10,edge.arrow.size=0.5,vertex.color=cols,vertex.label.cex=1,
     layout=layout.reingold.tilford(cc,root=seed))


### basic graph statistics
trl = mean(transitivity(cc,type='local',isolates='zero'))
apl = average.path.length(cc)
indeg = degree(cc,mode='in')
outdeg = degree(cc,mode='out')
ds = mean(cc_dat$dist)