Skip to content

Latest commit

 

History

History
172 lines (124 loc) · 5.17 KB

01_Data_Visuals.md

File metadata and controls

172 lines (124 loc) · 5.17 KB
Raw

NCRM SNA Short Course Project

School of Health and Wellbeing - UofG September, 2023

Teenage Friends and Lifestyle Data

The data was downloaded from the SIENA Data Repository under Open License Access, this also includes documentation for the Survey Design by West et al. (1997). The Glasgow Teenage data set consists of three waves, for the purpose of this task we look at a cross-sectional snapshot extracting Wave 2 outcomes.

01 - Dependencies

Load dependencies including that of library(statnet) to be able to parse data as Network objects. We also tap into data-processing libraries including library(dplyr).

# Pass the set of libraries we intend to load as a list
libraries <- c("statnet","dplyr","ggplot2","formatR")

wants <- libraries
has   <- wants %in% rownames(installed.packages())

# Adapted code to check if packages have already been installed
if(any(!has)) install.packages(wants[!has])

lapply(wants, library, character.only=TRUE, warn.conflicts=F, quietly=T)

Next load the survey files covering aspects from students’ lifestyle to friendship through the Teenage Friends and Lifestyle Dataset.

load("data/Glasgow-friendship.RData")
load("data/Glasgow-substances.RData")
load("data/Glasgow-geographic.RData")
load("data/Glasgow-demographic.RData")
load("data/Glasgow-geographic.RData")
load("data/Glasgow-lifestyle.RData")
load("data/Glasgow-various.RData")

02 - Read Network Data Structure

We map the elements of the Wave 2 friendship adjacency matrix onto a network object.

friendship_network <- as.network(as.matrix(friendship.2))

03 - Attach Vertex Attributes

Using the pipe operator %v% we assign vertex attributes to the network.

# Add a new attribute
friendship_network %v% "Smoking" <- c(tobacco[, 2])

Next extract an indicator of parent smoking from the family smoking data object. We recode the parent smoking variable solely for the purpose of attributing a color property to the network graph.

family_smoking <- as.numeric(familysmoking[, 2])
family_smoking <- case_match(family_smoking, c(1) ~ 4, c(2) ~
    5)

friendship_network %v% "Parent_Smoking" <- family_smoking

# Return a list of vertex attributes
list.vertex.attributes(friendship_network)
## [1] "na"             "Parent_Smoking" "Smoking"        "vertex.names"

04 - Attach Edge Attributes

Similar to using the vertex pipe operator, %v%, we now assign the edge operator, %e%, to include edge attributes to the network.

# Make an edge attribute
friendship_network %e% "Geographic Distance" <- as.matrix(distance.2)

list.edge.attributes(friendship_network)
## [1] "Geographic Distance" "na"

05 - Processing Network Data

After creating a vector containing arbitrary Student ID numbers, we map these on to corresponding vertices.

friend_names <- c(1:160)

network.vertex.names(friendship_network) <- friend_names

# Now change the shapes of the vertices in the network 4-
# stands for square, 3- stands for triangle
sex_gender_shapes <- as.numeric(sex.F)
sex_gender_shapes <- recode(sex_gender_shapes, `2` = 4, `1` = 3)

We can also look at a set of indicators summarising the properties of the friendship network and its attributes.

# Network attributes: vertices = 160 directed = TRUE hyper
# = FALSE loops = FALSE multiple = FALSE bipartite = FALSE
# total edges = 3910 missing edges = 906 non-missing edges
# = 3004 density = 0.1180818

# Vertex attributes:

# Parent_Smoking: numeric valued attribute attribute
# summary: Min. 1st Qu.  Median Mean 3rd Qu.  Max.  4.000
# 4.000 5.000 4.537 5.000 5.000

# Smoking: integer valued attribute 160 values
# vertex.names: character valued attribute 160 valid vertex
# names

# Edge attributes:

# Geographic Distance: numeric valued attribute attribute
# summary: Min. 1st Qu.  Median Mean 3rd Qu.  Max.  NA's
# 0.000 0.555 1.025 1.328 1.856 5.457 3537

06 - Graph of Smoking Behaviour Network

In order to plot a network graph of our data we use the plot.network() function weighing the length of edges according to students’ bee-line distance from each others’ homes making use of the edge.len option.

set.seed(11111)

plot.network(friendship_network, vertex.cex = "Smoking", vertex.col = "Parent_Smoking",
    vertex.sides = sex_gender_shapes, label.cex = 0.5, label = network.vertex.names(friendship_network),
    xlab = "Student ID", mode = "fruchtermanreingold", jitter = TRUE,
    displayisolates = TRUE, pad = 0.01, arrowhead.cex = 0.1,
    uselen = TRUE, edge.len = distance.2, edge.lwd = 0.01)

legend("bottomleft", fill = 4:5, legend = c("Yes", "No"), cex = 0.75,
    title = "Parent Smoking")

The network captures the following node attributes and related covariates:

  • Node Color - Parent smoking is captured by a dark blue hue.
  • Node Shape - Female and male sex is captured by triangle and square shape.
  • Node Size - Proportional to frequency with which a student smokes ranging [0,2].

We further note the importance of setting a seed to replicate the graph generation process using the fruchtermanreingold mode.