##1 network objects: importing, exploring, and manipulating network data
###########################################################
##1.1 Starting from scratch
##First, in Statnet:
library(statnet)

#Adding Edges
g <- network.initialize(5) # Create an empty graph
g[1,2] <- 1 # Add an edge from 1 to 2
g[2,] <- 1 # Add edges from 2 to everyone else
g # Examine the result

m <- matrix(0, nrow=5, ncol=5) # Create an adjacency matrix
m[3,4:5] <- 1 # Add entries from 3 to 4 and 5
g[m>0] <- 1 # Add more entries
g

#Delete edges
g[3,5] <- 0 # Remove the edge from 3 to 5
g # Its gone!
g[,] <- 0 # Remove all edges
g # Now, an empty graph

##Now, in igraph
library(igraph)

# The function make_graph() can work two ways:
# 1. as a vector of an even number of node IDs where each pair of adjacent nodes is considered an edge
# illustrated here with spacing
make_graph( c(1,2, 2,3, 3,4), directed=FALSE)
#or 2. using a symbolic formula
# `-` undirected tie
# `--+` directed tie (`+` is arrow's head)
# `:` refer to node sets (e.g. `A -- B:C` creates ties `A -- B` and `A -- C`)
# A network is either directed or undirected, it is not possible mix directed and undirected ties
g1 <- make_graph(~ A - B, B - C:D:E)
g2 <- make_graph(~ A --+ B, B +-- C, A --+ D:E, B --+ A)

g2
#To interpret the printout:
#First line includes
# - Class of the object (`IGRAPH`)
# - An ID of the object, not of particular interest (see also `?igraph::graph_id`)
#  A set of four slots for letter codes indicating, in order:
#  `U` or `D` if the network is `U`ndirected or `D`irected
#  `N` if the nodes have names
#  `W` if the network is weighted
#  `B` if the network is bipartite
#   Number of nodes
#   Number of edges
#   Starting with `+ attr:` list of present attributes, each of the form `nameoftheattribute (x/y)` where 
#   `x` informs about the type of an attribute: `v`ertex, `e`dge or `g`raph attribute
#   `y` informs about the mode of an attribute: `n`umeric, `c`haracter, `l`ogical, or e`x`tended (e.g. lists)
#   Starting with `+ edges:` a list of (some of) the edges of the network

###########################################
##2.2 Importing Relational Data
#Be sure to be in the directory where you stored the data for the workshop. If you've not set the working directory, you must do so now. See Section 1.7 for how to do this.
#You can use setwd() to change it, or platform specific shortcuts
getwd() # Check what directory you're in
list.files() # Check what's in the working directory

#Read an adjacency matrix (R stores it as a data frame by default)
relations <- read.csv("relationalData.csv",header=FALSE,stringsAsFactors=FALSE)
relations
#Here's a case where matrix format is preferred
relations <- as.matrix(relations) # convert to matrix format

#Read in some vertex attribute data (okay to leave it as a data frame)
nodeInfo <- read.csv("vertexAttributes.csv",header=TRUE)
nodeInfo

#Since our relational data has no row/column names, let's set them now
rownames(relations) <- nodeInfo$name
colnames(relations) <- nodeInfo$name
relations

#For more information. . .
?list.files
?read.csv
?as.matrix
?rownames

###########################################################
##2.3 Creating network objects

#throws an error if relations isn't a network!
nrelations<-network(as.matrix(relations),directed=FALSE) # Create a network object based on relations
nrelations # Get a quick description of the data

#now in igraph:
library(igraph)
irelations<-graph_from_adjacency_matrix(as.matrix(relations),mode="directed")

#For more information. . .
?network 
?as.network.matrix

###########################################################
##2.4 Description and Visualization

summary(irelations) #summary
ecount(irelations) #How many edges
vcount(irelations) #How many vertices

plot(irelations,
     layout=layout_with_fr,
     vertex.color="red",
     vertex.size=15,
     edge.arrow.size=0.5,
     vertex.label.color="black")

detach("package:igraph")

summary(nrelations) # Get an overall summary
network.dyadcount(nrelations) # How many dyads in nflo?
network.edgecount(nrelations) # How many edges are present?
network.size(nrelations) # How large is the network?
as.sociomatrix(nrelations) # Show it as a sociomatrix
nrelations[,] # Another way to do it

plot(nrelations,displaylabels=T) # Plot with names
plot(nrelations,displaylabels=T,mode="circle") # A less useful layout...

library(sna) # Load the sna library
gplot(nrelations) # Requires sna
gplot(relations) # gplot Will work with a matrix object too

#For more information. . .
?summary.network
?network.dyadcount
?network.edgecount
?as.sociomatrix
?as.matrix.network
?is.directed



###########################################################
##2.5 Network and Vertex Attributes 
#Add some attributes
nrelations %v% "id" <- nodeInfo$id # Add in our vertex attributes
nrelations %v% "age" <- nodeInfo$age
nrelations %v% "sex" <- nodeInfo$sex
nrelations %v% "handed" <- nodeInfo$handed
nrelations %v% "lastDocVisit" <- nodeInfo$lastDocVisit

#Listing attributes
list.vertex.attributes(nrelations) # List all vertex attributes
list.network.attributes(nrelations) # List all network attributes

#Retrieving attributes
nrelations %v% "age" # Retrieve vertex ages
nrelations %v% "id" # Retrieve vertex ids

#For more information. . .
?attribute.methods

####################################
#2.6 Edgelists

edgelist<-read.csv("edgeList.csv")
edgeNet<-network(edgelist,matrix.type="edgelist")
edgeNet

plot(edgeNet,displaylabels=T) #what's missing?

plot(edgeNet,displaylabels=T,edge.lwd=edgeNet%e%"Weight"*10)

##how can you extract the weights?
edgeNet[,]
edgeNet%e%"Weight"
as.sociomatrix.sna(edgeNet,"Weight")
as.edgelist.sna(edgeNet)

##now in igraph
library(igraph)

#this will throw an error - why?
iedge<-graph_from_edgelist(edgelist,directed=T)

iedge<-graph_from_data_frame(edgelist,directed=T)
plot(iedge) #where are the weights?

E(iedge)$Weight
plot(iedge,edge.width=E(iedge)$Weight*5)
detach("package:igraph")

###########################################################
###########################################################
##2 Classical Network Analysis with the SNA package
##2.1 Getting started
library(sna) # Load the sna library
library(help="sna") # See a list of help pages for sna package
load("introToSNAinR.Rdata")
# Load supplemental workshop data - note this name has changed from the video!

###########################################################
##2.2 Network visualization with gplot()
#Plotting the data we imported earlier
gplot(nrelations)
gplot(nrelations,displaylabels=TRUE) # This time with labels
#Let's color the nodes in sex-stereotypic colors
nodeColors<-ifelse(nodeInfo$sex=="F","hotpink","dodgerblue")
gplot(relations,gmode="graph",displaylabels=TRUE,vertex.col=nodeColors)

#Using data we just loaded in, plot the contiguity among nations in 1993 (from the Correlates of War (CoW)1 project)
gplot(contig_1993) # The default visualization
gplot(contig_1993, usearrows=FALSE) # Turn off arrows manually
gplot(contig_1993, gmode="graph") # Can also tell gplot the data is undirected

#We can add labels to the vertices
gplot(contig_1993, gmode="graph",displaylabels=TRUE,label.cex=0.5,label.col="blue")

#Other ways to specify the labeling
gplot(contig_1993,gmode="graph",label=network.vertex.names(contig_1993),label.cex=0.5,label.col="blue")

#Here's an example of directed data|militarized interstate disputes (MIDs) for 1993
gplot(mids_1993,label.cex=0.5,label.col="blue",displaylabels=TRUE)

#All those isolates can get in the way|we can suppress them using displayisolates
gplot(mids_1993,label.cex=0.5,label.col="blue",displaylabels=TRUE,displayisolates=FALSE)

#The default layout algorithm is that of Frutchterman-Reingold (1991), can use others
gplot(mids_1993,label.cex=0.5,label.col="blue",displaylabels=TRUE,displayisolates=FALSE,mode="circle") # The infamous circle

#or perhaps
gplot(mids_1993,label.cex=0.5,label.col="blue",displaylabels=TRUE,displayisolates=FALSE,mode="mds") # MDS of position similarity

#or perhaps
gplot(mids_1993,label.cex=0.5,label.col="blue",displaylabels=TRUE,displayisolates=FALSE,mode="target") # a target


#When a layout is generated, the results can be saved for later reuse:
coords <- gplot(contig_1993,gmode="graph",label=colnames(contig_1993[,]),label.cex=0.5,label.col="blue") # Capture the magic of the moment
coords # Show the vertex coordinates

#Saved (or a priori) layouts can be used via the coord argument
gplot(mids_1993,gmode="graph",label=colnames(contig_1993[,]),label.cex=0.5,label.col="blue",coord=coords)

#When the default settings are insuficient, interactive mode allows for tweaking
coords <- gplot(contig_1993, interactive=TRUE) # Modify and save
gplot(contig_1993,coord=coords,displaylabels=TRUE,gmode="graph",label.cex=0.5,label.col="blue") # Should reproduce the modified layout

#For more information. . .
?gplot ?gplot.layout