Specialization Programs Analysis.R

# Network Structure Analysis for Specializations

load("~/Documents/GitHub/SoftSkillsUniversityPrograms/PreProcessing.RData")
rm(list=setdiff(ls(), "Spec"))

library(igraph)
bn2 <- graph.data.frame(Spec,directed=FALSE)
bipartite.mapping(bn2)
V(bn2)$type <- bipartite_mapping(bn2)$type
V(bn2)$color <- ifelse(V(bn2)$type, "red", "green")
V(bn2)$shape <- ifelse(V(bn2)$type, "circle", "square")
#V(bn2)$label.cex <- ifelse(V(bn2)$type, 0.8, 1)
V(bn2)$size <- 3.5
E(bn2)$color <- "grey"
  plot(bn2, 
       vertex.label = NA, 
       layout = layout_nicely, 
       main = "")
  summary(bn2)
  V(bn2)$name[1:10]
  table(V(bn2)$type)
  V(bn2)$color[1:10]
  V(bn2)$shape[1:10]
  
  
  #library(rgexf)
  #pave <- igraph.to.gexf(bn2)
  #write.gexf(pave, output = "/home/jc/pave.gexf", replace = TRUE)
  
  # Network node prominence measures
  prominence <- data.frame(betweenness(bn2))
  prominence2 <- data.frame(degree(bn2))
  prominence3 <- data.frame(bonpow(bn2))
  prominence4 <- page.rank(bn2)
  prominence4 <- data.frame(prominence4$vector) 
  prominence4$Program <- rownames(prominence4)
  prominence5 <- data.frame(eigen_centrality(bn2))
  
  library(dplyr)
  HighestSkills <- prominence4 %>% filter(., prominence4$prominence4.vector > 0.0037)
  HS <- HighestSkills[69:93,]
  
  library(igraph)
  bn2 <- graph.data.frame(Spec,directed=FALSE)
  bipartite.mapping(bn2)
  V(bn2)$type <- bipartite_mapping(bn2)$type
  V(bn2)$color <- ifelse(V(bn2)$type, "red", "green")
  V(bn2)$shape <- ifelse(V(bn2)$type, "circle", "square")
  V(bn2)$label.cex <- ifelse(V(bn2)$type, 0.8, 1)
  V(bn2)$size <- sqrt(degree(bn2))
  E(bn2)$color <- "grey"
    plot(bn2, 
         vertex.label = NA, 
         layout = layout_components, 
         main = "")
    
    
    #TODAS2 <- TODAS %>% select(-from, -to, -pre, -post, -pattern) %>% left_join(aja, by = "docname") 
    
    
    table(degree(bn2,v=V(bn2)[type==FALSE]))
    mean(degree(bn2,v=V(bn2)[type==FALSE]))
    var(degree(bn2,v=V(bn2)[type==FALSE]))
    min(degree(bn2,v=V(bn2)[type==FALSE]))
    max(degree(bn2,v=V(bn2)[type==FALSE]))
    
    V(bn2)$deg <- degree(bn2)
    V(bn2)[type==FALSE & deg > 4]$name
    
    RelevantPrograms <- data.frame(cbind(
      Program = V(bn2)[type == FALSE]$name,
      Skills = V(bn2)[type==FALSE & deg >= 4]$deg))
    
    RelevantPrograms$Skills <- as.numeric(RelevantPrograms$Skills)
    
    
    ah <- data.frame(table(RelevantPrograms$Skills))
    colnames(ah)[1] <- "degree"
    ah$degree <- as.numeric(ah$degree)
    
    
    plot(ah$degree, ah$Freq, xlab = "degree", ylab= "Frequency")
    
    
    hist(RelevantPrograms$Skills, xlab= "Skills per program", main = "")
    
    bn2.pr <- bipartite.projection(bn2)
    Programs <- bn2.pr$proj1
    Terms <- bn2.pr$proj2
    c1 = cluster_fast_greedy(Terms)
    c2 = cluster_fast_greedy(Terms)
    
    # modularity measure
    modularity(c1)
    B = modularity_matrix(Programs, membership(c1))
    round(B[1,],2)
    membership(c1)
    length(c1)
    sizes(c1)
    crossing(c1, Terms)
    plot(c1, Terms, layout=layout_nicely(Terms, dim = 2))
    clique.number(Terms)
    largest_cliques(Terms)
    
    library(intergraph)
    graph.coreness(Terms)
    coreness <- graph.coreness(Terms)
    table(coreness)
    
    V(Terms)$color <- coreness + 1
    plot(Terms,vertex.label.cex=0.8, layout = layout_components)
    
    cluster_fast_greedy(Terms)
    cluster_louvain(Terms)
    cluster_spinglass(Terms)
    oye <- cluster_leading_eigen(Terms)
    oye2 <- cluster_leading_eigen(Programs)
    modularity(oye)
    modularity(oye2)
    plot_dendrogram(oye)
    plot_dendrogram(oye2)
    
    modularity(c2)
    B2 = modularity_matrix(Terms, membership(c2))
    round(B2[1,],2)
    membership(c2)
    length(c2)
    sizes(c2)
    crossing(c2, Terms)
    plot(c2, Terms, layout=layout_with_dh)
    plot(Terms, vertex.color=membership(c2), layout=layout_with_dh)
    
    graph.density(Programs)
    CentralityPrograms <- data.frame(degree(Programs))
    graph.density(Terms)
    get.adjacency(Programs)  
    T2 <- as.matrix(get.adjacency(Terms)  )
    
    summary(bn2)
    summary(Programs)
    graph.density(bn2)
    graph.density(Programs)
    graph.density(Terms)
    components(Programs)
    
    
    hum <- data.frame(degree(bn2))
    degree_distribution(bn2)
    hist(degree_distribution(bn2))
    ecount(bn2) #size of the graph (number of edges)
    edge.betweenness(bn2)
    hist(edge.betweenness(bn2))
    EB <- data.frame(edge.betweenness(bn2))
    
    edge.connectivity(bn2)
    edge.connectivity(bn2)
    SN <- as.matrix(get.adjacency(bn2))
    SN1 <- as.data.frame(SN)
    Skills <- as.data.frame(get.incidence(bn2))