# Network analysis
# Second session
# Harald Waxenecker



library('igraph')


### 1. Set the working directory

# using setwd()
setwd("/Users/macbookpro/Documents/")


# or: Session -> Set Working Directory -> Choose Directory


# Sample data: Florentine families
# https://methods-sagepub-com.ezproxy.muni.cz/dataset/intro-networks-in-florentine-1994
# https://sites.google.com/site/ucinetsoftware/datasets/padgett-florentine-families



### 2. Import data from different formats:

# .csv = comma separated values
# .xlsx = Excel file
# .Rdata = data format for use in R

# 2.1 Open Rdata with a click
# (here it is necessary to convert the names of the data objects)
nodes <- data
edges <- data

load("_binaries_dataset-florentine-1994-subset1-edges_dataset-florentine-1994-subset1-edges.Rdata")

# V(G)$name <- trimws(V(G)$name)


# 2.2 Import data from .csv
list.files()

nodes <-  read.csv("_binaries_dataset-florentine-1994-subset1-nodes_dataset-florentine-1994-subset1-nodes.csv")
edges <-  read.csv("_binaries_dataset-florentine-1994-subset1-edges_dataset-florentine-1994-subset1-edges.csv")

class(nodes)

# 2.3 Import data from .xlsx
install.packages("readxl")
library(readxl)

list.files()
nodes <- read_excel("_binaries_dataset-florentine-1994-subset1-nodes_dataset-florentine-1994-subset1-nodes.xlsx")
edges <- read_excel("_binaries_dataset-florentine-1994-subset1-edges_dataset-florentine-1994-subset1-edges.xlsx")

# In English, the Excel function "BUSCARV" is called "VLOOKUP".


### 3. Social network analysis

# igraph
# 3.1 The graph or network
florentine_network <- graph_from_data_frame(d=edges, vertices=nodes, directed=FALSE) 
class(florentine_network)

plot(florentine_network)


# node attributes:
V(florentine_network)$name = as.character(nodes$label)
V(florentine_network)$wealth

V(florentine_network)$totalties


# 3.2 General network characteristics

# one-mode network
# undirected network
edge_density(florentine_network)
ecount(florentine_network)
vcount(florentine_network)

diameter(florentine_network)
mean_distance(florentine_network)

distances(florentine_network, v = 'Medici', to = 'Peruzzi')
distances(florentine_network, v = 'Pazzi', to = 'Lamberteschi')
distances(florentine_network, v = 'Ginori', to = 'Strozzi')


# 3.3 Node centralities

degree(florentine_network)
betweenness(florentine_network, normalized = TRUE)
closeness(florentine_network, normalized = TRUE)


# Extract all vertex attributes into a data frame
node_centralities <- as_data_frame(florentine_network, what = "vertices")

# Compute centrality measures
deg  <- degree(florentine_network)
btw  <- betweenness(florentine_network, normalized = TRUE)
clo  <- closeness(florentine_network, normalized = TRUE)

# Add the computed centrality measures to the data frame
node_centralities$degree      <- deg
node_centralities$betweenness <- btw
node_centralities$closeness   <- clo


# 3.4 Improving the plot

# Plot the network with improved aesthetics

plot(florentine_network,
     layout = layout_with_fr(florentine_network),
     vertex.color = "gold",
     vertex.size = degree(florentine_network)*3+2,
     vertex.label.cex = degree(florentine_network)*0.5,
     main = "Florentine family network: nodesize = degree")


plot(florentine_network, 
     layout = layout_with_fr(florentine_network),              # using a different algorithm for node distribution
     vertex.label.cex = 1.2,                  # adjust label size
     vertex.size = degree(florentine_network)*5+1,             # define vertex size
     vertex.color = "gold",                   # define vertex color
     main = "Florentine family network: ")    # add a title



save.image(file = "exercise2_florentine_data.RData")


# Exercise:
# Create an edgelist and a nodelist in Excel. (>20 nodes)
# Import the data into R
# Create your network
# Do some node centrality analysis
# Plot your network