Maps
Adam Kiehl
2022-10-03
Data Source: The data for the below plots were
downloaded from the World University Ranking 2022-2023 open source data
set on Kaggle.com (https://www.kaggle.com/datasets/4c5d5aecdebdec38eee4f9de4546e92868be0f4dc1b0351078b307acec179158?resource=download).
World universities were scored and ranked based on their rankings in
four categories: education quality, post-graduation employability,
faculty quality, and research prestige.
Spatial Units: Countries were the spatial units chosen
for the plots below. University scores and category rankings were
aggregated by country to match this unit.
Purpose: The plots below were generated to convey a
sense of holistic higher education quality in countries across the
world. First and primarily, median university score per country was
plotted to give the best overall sense of a country’s collegiate
quality. Next, each of the four category rankings used to compute this
overall score (education quality, post-graduation employability, faculty
quality, and research prestige) were aggregated by country and plotted
as well. This serves to show how each country’s specific collegiate
strengths contribute to it’s overall score and may illuminate some
surprising aspects of the data.
Decisions: Several decisions were made in the process of
the plotting this university ranking data. First, since the spatial
composition of the map did not play a factor in the comparisons of
interest to this problem, a simple, aesthetically pleasing projection
similar to the Robinson projection was used, and a simple white theme
was used to maximize the data\(\colon\)ink ratio. Next, median was chosen
as a measure of central tendency over mean so that extremely prestigious
outlying universities (such as Harvard) were not overly influential.
Next, it was decided that each aggregated category ranking should be
plotted in addition to the aggregated overall university score, since
plotting different aspects of the data can tell different stories. For
these plots, the legend scales and color palettes were reversed because,
in these cases, a lower number is more desirable. It was additionally
decided that the four category plots should be plotted in a 2-by-2 frame
to emphasize their nature as categories contributing to a whole score.
Finally, an interactive hover effect was added to the main plot to
easily display the exact median university score for each country and to
allow users to zoom in on smaller countries.
# read in university ratings data
univ_data <- read.csv('university.csv') %>%
mutate(Education = as.numeric(Education.Rank),
Employ = as.numeric(Employability.Rank),
Faculty = as.numeric(Faculty.Rank),
Research = as.numeric(Research.Rank)) %>%
select(-c(Education.Rank,
Employability.Rank,
Faculty.Rank,
Research.Rank))
# fix unaligned regions between data frames
univ_data$Location[which(univ_data$Location == 'Northern Cyprus')] <- 'Cyprus'
univ_data$Location[which(univ_data$Location == 'Slovak Republic')] <- 'Slovakia'
univ_data$Location[which(univ_data$Location == 'United Kingdom')] <- 'UK'
head(univ_data)## World.Rank Institution Location National.Rank Score
## 1 1 Harvard University USA 1 100.0
## 2 2 Massachusetts Institute of Technology USA 2 96.7
## 3 3 Stanford University USA 3 95.1
## 4 4 University of Cambridge UK 1 94.1
## 5 5 University of Oxford UK 2 93.3
## 6 6 Princeton University USA 4 92.6
## Education Employ Faculty Research
## 1 1 1 1 1
## 2 4 12 2 7
## 3 11 4 3 2
## 4 3 25 4 10
## 5 7 27 9 4
## 6 5 15 6 79# read world map polygon data
world_map <- map_data('world') %>%
# create variable to preserve order
mutate(order = 1:n())# create aggregated scores data frame
univ_score <- univ_data %>%
# aggregate scores by country
group_by(Location) %>%
# compute median scores
summarize(Score = median(Score, na.rm = TRUE))
# left join world map data with university scores data
data1 <- world_map %>%
merge(univ_score,
by.x = 'region',
by.y = 'Location',
all.x = TRUE) %>%
# sort result by order variable
arrange(order)
# build median scores choropleth plot
plt1 <- ggplot(data1,
mapping = aes(x = long,
y = lat,
group = group,
# add choropleth data
fill = Score,
# add text aesthetic for hover effect
text = paste(region, ': ', round(Score, 1), sep = ''))) +
# draw map in white
geom_polygon(color = 'white', size = .05) +
# scale map to proper coordinates
coord_map(xlim = c(-200, 200),
ylim = c(-100, 100)) +
# choose a minimal theme
theme_dendro() +
# choose color palette
scale_fill_viridis(option = 'G') +
# add legend title
labs(fill = 'Median\nUniversity\nScore')
# add interactive hover effect
plt1 <- ggplotly(plt1, tooltip = 'text') %>%
# add plot title and subtitle
layout(title = list(text = paste0('<br>',
'Median University Rating by Country',
'<br>',
'<sup>',
'University rating based on education, employability, faculty, and research',
'<sup>')))
# display first plot
plt1# create aggregated ranks data frame
univ_edu <- univ_data %>%
# aggregate ranks by country
group_by(Location) %>%
# compute median education ranks
summarize(Education = median(Education, na.rm = TRUE))
# left join world map data with university education ranks
data2 <- world_map %>%
merge(univ_edu,
by.x = 'region',
by.y = 'Location',
all.x = TRUE) %>%
# sort result by order variable
arrange(order)
# build median education ranks choropleth plot
plt2 <- ggplot(data2,
mapping = aes(x = long,
y = lat,
group = group,
# add choropleth data
fill = Education)) +
# draw map in white
geom_polygon(color = 'white', size = .05) +
# scale map to proper coordinates
coord_map(xlim = c(-200, 200),
ylim = c(-100, 100)) +
# choose a minimal theme
theme_dendro() +
# choose color palette
scale_fill_viridis(option = 'F', direction = -1) +
# reverse order of legend scale
guides(fill = guide_legend(reverse = FALSE)) +
# add title and legend title
labs(title = 'Education',
fill = 'Median\nUniversity\nEducation\nRank')# create aggregated ranks data frame
univ_employ <- univ_data %>%
# aggregate ranks by country
group_by(Location) %>%
# compute median employability ranks
summarize(Employ = median(Employ, na.rm = TRUE))
# left join world map data with university employability ranks
data3 <- world_map %>%
merge(univ_employ,
by.x = 'region',
by.y = 'Location',
all.x = TRUE) %>%
# sort result by order variable
arrange(order)
# build median employability ranks choropleth plot
plt3 <- ggplot(data3,
mapping = aes(x = long,
y = lat,
group = group,
# add choropleth data
fill = Employ)) +
# draw map in white
geom_polygon(color = 'white', size = .05) +
# scale map to proper coordinates
coord_map(xlim = c(-200, 200),
ylim = c(-100, 100)) +
# choose a minimal theme
theme_dendro() +
# choose color palette
scale_fill_viridis(option = 'F', direction = -1) +
# reverse order of legend scale
guides(fill = guide_legend(reverse = FALSE)) +
# add title and legend title
labs(title = 'Employability',
fill = 'Median\nUniversity\nEmployability\nRank')# create aggregated ranks data frame
univ_faculty <- univ_data %>%
# aggregate ranks by country
group_by(Location) %>%
# compute median faculty ranks
summarize(Faculty = median(Faculty, na.rm = TRUE))
# left join world map data with university faculty ranks
data4 <- world_map %>%
merge(univ_faculty,
by.x = 'region',
by.y = 'Location',
all.x = TRUE) %>%
# sort result by order variable
arrange(order)
# build median faculty ranks choropleth plot
plt4 <- ggplot(data4,
mapping = aes(x = long,
y = lat,
group = group,
# add choropleth data
fill = Faculty)) +
# draw map in white
geom_polygon(color = 'white', size = .05) +
# scale map to proper coordinates
coord_map(xlim = c(-200, 200),
ylim = c(-100, 100)) +
# choose a minimal theme
theme_dendro() +
# choose color palette
scale_fill_viridis(option = 'F', direction = -1) +
# reverse order of legend scale
guides(fill = guide_legend(reverse = FALSE)) +
# add title and legend title
labs(title = 'Faculty',
fill = 'Median\nUniversity\nFaculty\nRank')# create aggregated ranks data frame
univ_research <- univ_data %>%
# aggregate ranks by country
group_by(Location) %>%
# compute median faculty ranks
summarize(Research = median(Research, na.rm = TRUE))
# left join world map data with university research ranks
data5 <- world_map %>%
merge(univ_research,
by.x = 'region',
by.y = 'Location',
all.x = TRUE) %>%
# sort result by order variable
arrange(order)
# build median research ranks choropleth plot
plt5 <- ggplot(data5,
mapping = aes(x = long,
y = lat,
group = group,
# add choropleth data
fill = Research)) +
# draw map in white
geom_polygon(color = 'white', size = .05) +
# scale map to proper coordinates
coord_map(xlim = c(-200, 200),
ylim = c(-100, 100)) +
# choose a minimal theme
theme_dendro() +
# choose color palette
scale_fill_viridis(option = 'F', direction = -1) +
# reverse order of legend scale
guides(fill = guide_legend(reverse = FALSE)) +
# add title and legend title
labs(title = 'Research',
fill = 'Median\nUniversity\nResearch\nRank')# display category plots
grid.arrange(plt2, plt3, plt4, plt5,
# define plotting grid
ncol = 2,
# add title to grid
top = 'Median Category Rankings')