Advanced Data Visualization in R

Iris Malone

November 6, 2015

Iris Malone Advanced Data Visualization in R November 6, 2015 1 / 68

Outline

What I’m Covering:

Iris Malone Advanced Data Visualization in R November 6, 2015 2 / 68

Outline

What I’m Covering:

pros/cons different graphing packages

Iris Malone Advanced Data Visualization in R November 6, 2015 2 / 68

Outline

What I’m Covering:

pros/cons different graphing packages

ggplot and the grammar of graphics

Iris Malone Advanced Data Visualization in R November 6, 2015 2 / 68

Outline

What I’m Covering:

pros/cons different graphing packages

ggplot and the grammar of graphics
how to visualize summary stats and regression results

Iris Malone Advanced Data Visualization in R November 6, 2015 2 / 68

Outline

What I’m Covering:

pros/cons different graphing packages

ggplot and the grammar of graphics
how to visualize summary stats and regression results
basic spatial visualization

Iris Malone Advanced Data Visualization in R November 6, 2015 2 / 68

Download code and slides at:

web.stanford.edu/~imalone/VAM.html

Iris Malone Advanced Data Visualization in R November 6, 2015 3 / 68

Choosing a Visualization Package

Toy Example: Plot number of democracies and autocracies over time.

Iris Malone Advanced Data Visualization in R November 6, 2015 4 / 68

Choosing a Visualization Package

Toy Example: Plot number of democracies and autocracies over time.

Options:

plot (graphics)

Iris Malone Advanced Data Visualization in R November 6, 2015 4 / 68

Option 1: Plot
Code:

plot(year, numdems, #x, y

#aesthetics stuff (color, plot type, size)
col = "blue", type = "l" , lwd=3,
#main title and axes labels
main = "Number of Democracies and Autocracies, 1800-2012",
xlab = "Year",
ylab = "Count")
lines(year, numauts, type = "l", col = "red", lwd =3)
legend("topleft", #location of legend
legend=c("Democracies","Autocracies"), #legend labels/var
col = c(4, 2), # colors for each,
#legend colors default 1= black, 2 = red, 4=blue
lty = c(1, 1),
title="Regime Type") # Name of Legend

Iris Malone Advanced Data Visualization in R November 6, 2015 5 / 68

Option 1: Plot
Visual:
Number of Democracies and Autocracies, 1800−2012

Regime Type
Democracies
Autocracies
80
60
Count

40
20
0

1800 1850 1900 1950 2000

Year

Iris Malone Advanced Data Visualization in R November 6, 2015 6 / 68

Choosing a Visualization Package

Toy Example.
Options:

plot (graphics)

Iris Malone Advanced Data Visualization in R November 6, 2015 7 / 68

Choosing a Visualization Package

Toy Example.
Options:

plot (graphics) Overly simplistic.

Iris Malone Advanced Data Visualization in R November 6, 2015 7 / 68

Choosing a Visualization Package

Toy Example.
Options:

plot (graphics) Overly simplistic. Error interp.

Iris Malone Advanced Data Visualization in R November 6, 2015 7 / 68

Choosing a Visualization Package

Toy Example.
Options:

plot (graphics) Overly simplistic. Error interp. Limited online support.

Iris Malone Advanced Data Visualization in R November 6, 2015 7 / 68

Choosing a Visualization Package

Toy Example.
Options:

plot (graphics) Overly simplistic. Error interp. Limited online support.

xyplot (lattice)

Iris Malone Advanced Data Visualization in R November 6, 2015 7 / 68

Option 2: xyplot

Step 1: Build First Layer for Number of Democracies.

library(lattice)
layer1 = xyplot(numdems ~ year, #format: y ~ x
type = "l",
# add title
main = "Number of Democracies and Autocracies, 1800-2012",
col = "blue", lwd=1,
#legend
key=list(space="right", # location
# aesthetics (aes)
lines=list(col=c("red","blue"), lty=c(1,1), lwd=2),
#labels for each line
text=list(c("Autocracies","Democracies"))))

Iris Malone Advanced Data Visualization in R November 6, 2015 8 / 68

Option 2: xyplot

Step 2: Build Second Layer for Number of Autocracies.

layer2 = xyplot(numauts ~ year,

type = "l",
col = "red")

Iris Malone Advanced Data Visualization in R November 6, 2015 9 / 68

Option 2: xyplot
Step 3: Add the layers.

#need extra package to put layers on top

suppressMessages(library(latticeExtra))
layer1 + layer2

Number of Democracies and Autocracies, 1800−2012

100

80

60
numdems

Autocracies
Democracies
40

20

1800 1850 1900 1950 2000

year

Iris Malone Advanced Data Visualization in R November 6, 2015 10 / 68

Graphics Packages

Options:

plot (graphics) Overly simplistic. Error Interp. Limited online support.

xyplot (lattice)

Iris Malone Advanced Data Visualization in R November 6, 2015 11 / 68

Graphics Packages

Options:

plot (graphics) Overly simplistic. Error Interp. Limited online support.

xyplot (lattice) Supplemental packages.

Iris Malone Advanced Data Visualization in R November 6, 2015 11 / 68

Graphics Packages

Options:

plot (graphics) Overly simplistic. Error Interp. Limited online support.

xyplot (lattice) Supplemental packages. Default look could be nicer.

Iris Malone Advanced Data Visualization in R November 6, 2015 11 / 68

Graphics Packages

Options:

plot (graphics) Overly simplistic. Error Interp. Limited online support.

xyplot (lattice) Supplemental packages. Default look could be nicer.
ggplot (ggplot2)

Iris Malone Advanced Data Visualization in R November 6, 2015 11 / 68

Option 3: ggplot

suppressMessages(library(ggplot2))
ggplot(data = NULL) +
#line for dems and auts
geom_line(aes(x=year, y = numdems, colour = "numdems")) +
geom_line(aes(x=year, y = numauts, colour = "numauts")) +
xlab("Year") + ylab("Count") +
ggtitle("Number of Democracies and Autocracies, 1800-2012") +
# legend aesthetics
scale_color_manual(name = "Regime Type", # Name
labels = c(numdems="Democracies", numauts = "Autocracies"), #
values=c(numdems=4,numauts=2))

Iris Malone Advanced Data Visualization in R November 6, 2015 12 / 68

Option 3: ggplot

Number of Democracies and Autocracies, 1800−2012

100

75

Regime Type
Count

50 Autocracies
Democracies

25

1800 1850 1900 1950 2000

Year

Iris Malone Advanced Data Visualization in R November 6, 2015 13 / 68

Summary of Options:

plot (graphics)
xyplot (lattice)
ggplot (ggplot2)

Iris Malone Advanced Data Visualization in R November 6, 2015 14 / 68

Why ggplot?

Used professionally

Iris Malone Advanced Data Visualization in R November 6, 2015 15 / 68

Why ggplot?
Example 1

Iris Malone Advanced Data Visualization in R November 6, 2015 16 / 68

Why ggplot?
Example 2

Iris Malone Advanced Data Visualization in R November 6, 2015 17 / 68

Why ggplot?

- Used professionally

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty
- Easy to manipulate

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty
- Easy to manipulate
- Great support online

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty
- Easy to manipulate
- Great support online
- Knowledge transfers to
other packages/languages
(ggvis, Shiny, Python)

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty
- Easy to manipulate
- Great support online
- Knowledge transfers to
other packages/languages
(ggvis, Shiny, Python)
- Steep Learning Curve

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty
- Easy to manipulate
- Great support online
- Knowledge transfers to
other packages/languages
(ggvis, Shiny, Python)
- Steep Learning Curve
- Lots of syntax

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty
- Easy to manipulate
- Great support online
- Knowledge transfers to
other packages/languages
(ggvis, Shiny, Python)
- Steep Learning Curve
- Lots of syntax
- Can be slow

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty
- Easy to manipulate
- Great support online
- Knowledge transfers to
other packages/languages
(ggvis, Shiny, Python)
- Steep Learning Curve
- Lots of syntax
- Can be slow
- Defaults to weird colors

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

Why ggplot?

- Used professionally
- Very pretty
- Easy to manipulate
- Great support online
- Knowledge transfers to
other packages/languages
(ggvis, Shiny, Python)
- Steep Learning Curve
- Lots of syntax
- Can be slow
- Defaults to weird colors
- Summary: Worth it.

Iris Malone Advanced Data Visualization in R November 6, 2015 18 / 68

ggplot Syntax

Following next few slides adapated from Samantha Tyner

Based on Grammar of Graphics book by Leland Wilkinson hence ‘gg’

Iris Malone Advanced Data Visualization in R November 6, 2015 19 / 68

ggplot Syntax

Following next few slides adapated from Samantha Tyner

Based on Grammar of Graphics book by Leland Wilkinson hence ‘gg’

New Zealander Hadley Wickham → R

Iris Malone Advanced Data Visualization in R November 6, 2015 19 / 68

ggplot Syntax

Following next few slides adapated from Samantha Tyner

Based on Grammar of Graphics book by Leland Wilkinson hence ‘gg’

New Zealander Hadley Wickham → R
Analogy: Think of parts of a plot like parts of a sentence

Iris Malone Advanced Data Visualization in R November 6, 2015 19 / 68

ggplot Syntax

Following next few slides adapated from Samantha Tyner

Based on Grammar of Graphics book by Leland Wilkinson hence ‘gg’

New Zealander Hadley Wickham → R
Analogy: Think of parts of a plot like parts of a sentence
Warning: qplot

Iris Malone Advanced Data Visualization in R November 6, 2015 19 / 68

ggplot Syntax
Noun → Data

ggplot(data = df)

Iris Malone Advanced Data Visualization in R November 6, 2015 20 / 68

ggplot Syntax
Noun → Data

ggplot(data = df)

Verb → “geom_” + Plot Type

ggplot(data = df) + geom_bar()

Table 1: Geom Types

abline area boxplot

errorbar histogram line
point ribbon smooth
blank density jitter
polygon quantile vline

Iris Malone Advanced Data Visualization in R November 6, 2015 20 / 68

ggplot Syntax

Noun → Data

ggplot(data = df)

Verb → “geom_” + Plot Type

ggplot(data = df) + geom_bar()

Adjectives → Aesthetics (“aes”) (x, y, fill, colour, linetype)

ggplot(data = df, aes(x=categorical.var, fill=group.var)) +

geom_bar()

Iris Malone Advanced Data Visualization in R November 6, 2015 21 / 68

ggplot Syntax

Adjectives → Aesthetics (“aes”)

Iris Malone Advanced Data Visualization in R November 6, 2015 22 / 68

ggplot Syntax. Aesthetics Sidebar.
Note. Difference between fill, colour, and placement.
Default.

ggplot(data = NULL, aes(x=numdems)) + geom_bar()

## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x

40
count

20

0 25 50 75 100
numdems

Iris Malone Advanced Data Visualization in R November 6, 2015 23 / 68

ggplot Syntax. Aesthetics Sidebar.
Fill.

ggplot(data = NULL, aes(x=numdems)) +

geom_bar(fill="red")

## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x

40
count

20

0 25 50 75 100
numdems

Iris Malone Advanced Data Visualization in R November 6, 2015 24 / 68

ggplot Syntax. Aesthetics Sidebar.
Colour.

ggplot(data = NULL, aes(x=numdems)) +

geom_bar(colour="red")

## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x

40
count

20

0 25 50 75 100
numdems

Iris Malone Advanced Data Visualization in R November 6, 2015 25 / 68

ggplot Syntax. Aesthetics Sidebar.
Fill and Colour.

ggplot(data = NULL, aes(x=numdems)) +

geom_bar(fill="white", colour="red")

## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x

40
count

20

0 25 50 75 100
numdems

Iris Malone Advanced Data Visualization in R November 6, 2015 26 / 68

ggplot Syntax. Aesthetics Sidebar
Defined inside the aesthetics argument. Ack!

ggplot(data = NULL) +
geom_bar(aes(x=numdems, colour ="red"))

## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x

40
count

"red"
red

20

0 25 50 75 100
numdems

Iris Malone Advanced Data Visualization in R November 6, 2015 27 / 68

ggplot Syntax. Aesthetics Sidebar
Defined inside the aesthetics argument.

ggplot(data = NULL) +
geom_bar(aes(x=numdems,
fill=factor(I(year)<1950)))

## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x

40

factor(I(year) < 1950)

count

FALSE
TRUE

20

0 25 50 75
Iris Malone Advancednumdems
Data Visualization in R November 100
6, 2015 28 / 68
ggplot Syntax

Adverb → stat (e.g. identity, bin)

ggplot(data = df, aes(x=categorical.var, fill=group.var))

+ geom_bar(stat = "bin")

Iris Malone Advanced Data Visualization in R November 6, 2015 29 / 68

ggplot Syntax

Adverb → stat (e.g. identity, bin)

ggplot(data = df, aes(x=categorical.var, fill=group.var))

+ geom_bar(stat = "bin")

Preposition → position (e.g. fill, dodge, identity)

ggplot(data = df, aes(x=categorical.var, fill=group.var)) +

geom_bar(stat="bin", position = "identity", binwidth=5)

Iris Malone Advanced Data Visualization in R November 6, 2015 29 / 68

Toy Example for Learning Syntax.

Fearon and Laitin (2003). Let’s suppose we would like to get a feel for their
data by first just looking at the number of civil wars in their dataset as a
function of two variables: (1) region and (2) time.

library(foreign)
df = read.dta("repdata.dta")
#subset data so 1 obs/civil war
dfonset = subset(df, df$onset == 1)

D.V. Civil War Onset

I.V. Ethnic Fractionalization, Mountainous Terrain, Oil, New State, Others

Iris Malone Advanced Data Visualization in R November 6, 2015 30 / 68

ggplot Syntax. Example. Civil War by Region and Decade

p = ggplot(data = dfonset, aes(x=decade,

fill = region)) +
geom_bar(position = "identity",
binwidth=5)
p

10

region
western democracies and japan
e. europe and the former soviet union
count

asia
n. africa and the middle east
sub−saharan africa
5 latin america and the caribbean

1960 1970 1980 1990

decade

Iris Malone Advanced Data Visualization in R November 6, 2015 31 / 68

ggplot Syntax. Example. Civil War by Region and Decade

p = ggplot(data = dfonset, aes(x=decade,

fill = region)) +
geom_bar(position = "dodge",
binwidth=5)
p

10

region
western democracies and japan
e. europe and the former soviet union
count

asia
n. africa and the middle east
sub−saharan africa
5 latin america and the caribbean

1960 1970
decade
1980 1990
5
Iris Malone Advanced Data Visualization in R November 6, 2015 32 / 68
ggplot Syntax: Title
Add Title.

p = p +
ggtitle("Civil Wars by Space and Time")
p

Civil Wars by Space and Time

10
region
western democracies and japan
e. europe and the former soviet union
count

asia
n. africa and the middle east
sub−saharan africa
5 latin america and the caribbean

1960 1970 1980 1990

decade

Iris Malone Advanced Data Visualization in R November 6, 2015 33 / 68

ggplot Syntax: Axes
Add Axes.

p = p + xlab("Decade") +
ylab("Civil War Frequency")
p

Civil Wars by Space and Time

10
region
Civil War Frequency

western democracies and japan

e. europe and the former soviet union
asia
n. africa and the middle east
sub−saharan africa
5 latin america and the caribbean

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 34 / 68

ggplot Syntax: Theme

Change Theme.

p = p + theme_bw()
p

Civil Wars by Space and Time

10
region
Civil War Frequency

western democracies and japan

e. europe and the former soviet union
asia
n. africa and the middle east
sub−saharan africa
5 latin america and the caribbean

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 35 / 68

ggplot Syntax: Theme

Change Theme.

p = p + theme_classic() #looks like plot!

p

Civil Wars by Space and Time

10
region
Civil War Frequency

western democracies and japan

e. europe and the former soviet union
asia
n. africa and the middle east
sub−saharan africa
5 latin america and the caribbean

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 36 / 68

ggplot Syntax: Theme

What’s great about ggplot is you can customize your own! This is what’s
going on, under the hood, for theme_classic for example:

p = p + theme(panel.grid.major = element_blank(), #grid lines

panel.grid.minor = element_blank() ,
panel.border = element_blank(), #border
panel.background = element_blank(), #background
#change axes line
axis.line = element_line(colour = "black"),
axis.text.x=element_text(colour="black"),
axis.text.y=element_text(colour="black"))

Iris Malone Advanced Data Visualization in R November 6, 2015 37 / 68

ggplot Syntax: Color.
Change Colors. Reds! Use scale_fill_manual for bins,
scale_colour_manual for lines, points, and scale_linetype_manual
otherwise.

rhg_cols = c("#771C19","#AA3929","#E25033","#F27314",
"#F8A31B","#E2C59F","#556670","#000000")
p = p + scale_fill_manual(values = rhg_cols)
p
Civil Wars by Space and Time

10 region
Civil War Frequency

western democracies and japan

e. europe and the former soviet union
asia
n. africa and the middle east
sub−saharan africa
5
latin america and the caribbean

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 38 / 68

ggplot Syntax: Color.
Or choose others! Blues!
#default brewer colors
p = p +
scale_fill_brewer()

## Scale for 'fill' is already present. Adding another scale f

p
Civil Wars by Space and Time

10 region
Civil War Frequency

western democracies and japan

e. europe and the former soviet union
asia
n. africa and the middle east
sub−saharan africa
5
latin america and the caribbean

1960 1970 1980 1990

Decade
Iris Malone Advanced Data Visualization in R November 6, 2015 39 / 68
ggplot Syntax: Color.
#default grey
p = p + scale_fill_grey()

## Scale for 'fill' is already present. Adding another scale f

p
Civil Wars by Space and Time

10
region
Civil War Frequency

western democracies and japan

e. europe and the former soviet union
asia
n. africa and the middle east
sub−saharan africa
5 latin america and the caribbean

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 40 / 68

ggplot Syntax: Legend.

Change Legend Labels (or Colors)

p = p +
scale_fill_manual(name ="My Super Awesome Legend Title",
values=c("darkblue", "darkred", "darkgreen",
"grey", "darkorange", "purple"),
labels=c("Western Dems and Japan", "Former USSR", "Asia",
"North Africa and \n Middle East", "Sub-Saharan Africa",
"Latin America"))

## Scale for 'fill' is already present. Adding another scale f

Iris Malone Advanced Data Visualization in R November 6, 2015 41 / 68

ggplot Syntax: Legend.

Why define fill (or color or linetype) inside aes? Keep track of variables!
Recall:

ggplot(data = NULL, aes(x = year)) +

geom_line(aes(y = numdems, colour = "numdems")) +
geom_line(aes(y = numauts, colour = "numauts")) +
scale_color_manual(name = "Regime Type",
labels = c(numdems="Democracies", numauts = "Autocracies")
values=c(numdems="blue",numauts="red"))

Iris Malone Advanced Data Visualization in R November 6, 2015 42 / 68

ggplot Syntax: Legend.

Note: p is ggplot object with 2 components in aes: x = decade and fill =

region

p = p +
scale_fill_manual(name ="My Super Awesome Legend Title",
values=c("darkblue", "darkred", "darkgreen",
"grey", "darkorange", "purple"),
labels=c("Western Dems and Japan", "Former USSR", "Asia",
"North Africa and \n Middle East", "Sub-Saharan Africa",
"Latin America"))

## Scale for 'fill' is already present. Adding another scale f

Iris Malone Advanced Data Visualization in R November 6, 2015 43 / 68

ggplot Syntax: Legend

Civil Wars by Space and Time

10
My Super Awesome Legend Title
Civil War Frequency

Western Dems and Japan

Former USSR
Asia
North Africa and
Middle East
Sub−Saharan Africa
5 Latin America

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 44 / 68

ggplot Syntax: Legend.

Change the position of the legend.

p = p + theme(legend.position="top")
p

Civil Wars by Space and Time

North Africa and
My Super Awesome Legend Title Western Dems and Japan Former USSR Asia Sub−Saharan Africa Latin America
Middle East

10
Civil War Frequency

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 45 / 68

ggplot Syntax: Legend.

Change the position of the legend.

# Position legend in graph, where x,y is 0,0 (bottom left)

# to 1,1 (top right)
p = p + theme(legend.position=c(0.10, .8))
p

Civil Wars by Space and Time

My Super Awesome Legend Title
Western Dems and Japan
Former USSR
Asia
North Africa and
Middle East
10 Sub−Saharan Africa
Civil War Frequency

Latin America

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 46 / 68

ggplot Syntax: Legend.

Remove the legend.

p = p + theme(legend.position="none")
p

Civil Wars by Space and Time

10
Civil War Frequency

1960 1970 1980 1990

Decade

Iris Malone Advanced Data Visualization in R November 6, 2015 47 / 68

ggplot Syntax: Annotation.
Suppose you want to add a label. For example, what’s the one western
democracy with a civil war in the 1960s?
#it's the UK vs the IRA
p = p + annotate("text", label = "UK vs IRA",
x = 1959, y = 2, size = 6, colour = "black")
p
Civil Wars by Space and Time

10
Civil War Frequency

UK vs IRA

1950 1960 1970 1980 1990

Decade
Iris Malone Advanced Data Visualization in R November 6, 2015 48 / 68
ggplot Syntax: Multiple plots.
Suppose you want a separate barplot for every decade.

ggplot(data = dfonset, aes(x=decade, group = region,

fill = factor(region))) +
geom_bar(stat="bin", position = "dodge", binwidth=5) +
facet_wrap(~region) #var you want separate plots by
western democracies and japan e. europe and the former soviet union asia

10

5
factor(region)
western democracies and japan

0 e. europe and the former soviet union

count

n. africa and the middle east sub−saharan africa latin america and the caribbean asia
n. africa and the middle east
sub−saharan africa
latin america and the caribbean
10

0
1960 1970 1980 1990 1960 1970 1980 1990 1960 1970 1980 1990
decade

Iris Malone Advanced Data Visualization in R November 6, 2015 49 / 68

ggplot Syntax: Saving results.

Option 1.

pdf("nameoffile.pdf", width=12, height = 5)

p
dev.off()

## pdf
## 2

Option 2.

ggsave(p, file="nameoffile.pdf", width=12, height=5)

Iris Malone Advanced Data Visualization in R November 6, 2015 50 / 68

Visualizing Regressions

Visualizing Regression Results

Coef Plots
Not covered here: marginal effects, predicted outcomes

Iris Malone Advanced Data Visualization in R November 6, 2015 51 / 68

Coef Plot: Canned Function

m1 = glm(onset ~ warl + gdpenl + lpopl1 + lmtnest + ncontig +

library(coefplot)
coefplot(m1)

Coefficient Plot
relfrac

ethfrac

polity2l

instab

nwstate
Coefficient

Oil

ncontig

lmtnest

lpopl1

gdpenl

warl

(Intercept)

−6 −3 0 3
Value

Iris Malone Advanced Data Visualization in R November 6, 2015 52 / 68

Coefplot: Our Own Function
Adapted from Stat Bandit
#Format the data
coefplot.gg = function(model, data){
# data is a data frame with 4 columns
# data$names gives variable names
# data$modelcoef gives center point
# data$ylo gives lower limits
# data$yhi gives upper limits
modelcoef = summary(model)$coefficients[1:length(model$coeff
modelse = summary(model)$coefficients[1:length(model$coeffic
ylo = modelcoef - qt(.975, nrow(data))*(modelse)
yhi = modelcoef + qt(.975, nrow(data))*(modelse)
names = names(m1$coefficients)
dfplot = data.frame(names, modelcoef, modelse, ylo, yhi)
# ...
}
Iris Malone Advanced Data Visualization in R November 6, 2015 53 / 68
Coefplot
Define the plot

coefplot.gg = function(model, data){

# ...
#define plot
library(ggplot2)
p = ggplot(dfplot, aes(x=names,
y=modelcoef,
ymin=ylo, ymax=yhi))
+ geom_pointrange(colour=ifelse(ylo < 0 & yhi > 0,
"red", "blue"))
+ theme_bw() + coord_flip()
+ geom_hline(aes(x=0), lty=2)
+ xlab('Variable') + ylab('')
return(p)
}

Iris Malone Advanced Data Visualization in R November 6, 2015 54 / 68

Coefplot
Evaluate the function

coefplot.gg(m1, df)

warl

relfrac

polity2l

Oil

nwstate

ncontig
Variable

lpopl1

lmtnest

instab

gdpenl

ethfrac

(Intercept)

−6 −3 0 3

Iris Malone Advanced Data Visualization in R November 6, 2015 55 / 68

Spatial Visualization

Packages:

rworldmap
maps
ggmap

suppressMessages(library(maps))
suppressMessages(library(ggmap))
suppressMessages(library(mapproj))

Iris Malone Advanced Data Visualization in R November 6, 2015 56 / 68

maps

Adapted from Mahbubul Majumder.

dfworldmap = map_data("world")
ggplot() + geom_polygon(aes(x=long,y=lat, group=group),
fill="grey65",
data=dfworldmap) + theme_bw()

50
lat

−50

−100 0 100 200

long

Iris Malone Advanced Data Visualization in R November 6, 2015 57 / 68

Chloropleth maps
Suppose we want to map different levels of a variable by some unit like a
state or country. For a toy example, we’ll map 1973 murder rates by state
using the USArrests data.
Step 1: Format data

suppressMessages(library(dplyr))
us = map_data("state")
head(us)

## long lat group order region subregion

## 1 -87.46201 30.38968 1 1 alabama <NA>
## 2 -87.48493 30.37249 1 2 alabama <NA>
## 3 -87.52503 30.37249 1 3 alabama <NA>
## 4 -87.53076 30.33239 1 4 alabama <NA>
## 5 -87.57087 30.32665 1 5 alabama <NA>
## 6 -87.58806 30.32665 1 6 alabama <NA>
Iris Malone Advanced Data Visualization in R November 6, 2015 58 / 68
Chloropleth maps

head(USArrests)

## Murder Assault UrbanPop Rape

## Alabama 13.2 236 58 21.2
## Alaska 10.0 263 48 44.5
## Arizona 8.1 294 80 31.0
## Arkansas 8.8 190 50 19.5
## California 9.0 276 91 40.6
## Colorado 7.9 204 78 38.7

#mismatch between region and state

# need to add var to arrest data to match 1:1

Iris Malone Advanced Data Visualization in R November 6, 2015 59 / 68

Chloropleth maps
arrest = USArrests %>%
add_rownames("region") %>%
#use mutate function from plyr
mutate(region=tolower(region)) #make it all lowercase
#format to work with map
head(arrest)

## Source: local data frame [6 x 5]

##
## region Murder Assault UrbanPop Rape
## (chr) (dbl) (int) (int) (dbl)
## 1 alabama 13.2 236 58 21.2
## 2 alaska 10.0 263 48 44.5
## 3 arizona 8.1 294 80 31.0
## 4 arkansas 8.8 190 50 19.5
## 5 california 9.0 276 91 40.6
## 6 colorado
Iris Malone
7.9Advanced Data
204Visualization in R78 38.7 November 6, 2015 60 / 68
Chloropleth maps
Step 2: Plot the base map layer

g = ggplot()
#must define map first
g = g + geom_map(data=us, map=us,
aes(x=long, y=lat, map_id=region),
fill="#ffffff", color="#ffffff", size=0.15
g
50

45

40
lat

35

30

25

−120 −100 −80

long

Iris Malone Advanced Data Visualization in R November 6, 2015 61 / 68

Chloropleth maps

Step 3: Add our arrest data

g = g + geom_map(data=arrest, map=us,
aes(fill=Murder, map_id=region),
color="#ffffff", size=0.15)
g

50

45

Murder

40 15
lat

10
35
5

30

25

−120 −100 −80

long

Iris Malone Advanced Data Visualization in R November 6, 2015 62 / 68

Chloropleth maps
Step 4: Make it look pretty.

g = g + scale_fill_continuous(low='thistle2', high='darkblue',
guide='colorbar') + xlab("")
g = g + theme(panel.border = element_blank()) + theme(panel.ba
g

Murder

15

10

Iris Malone Advanced Data Visualization in R November 6, 2015 63 / 68

ggmap
map1 = suppressMessages(get_map(
location = 'Stanford University', zoom = 14, #zoom-in level
maptype="satellite")) #map type
ggmap(map1)

37.44

37.43
lat

37.42

37.41
Iris Malone Advanced Data Visualization in R November 6, 2015 64 / 68
ggmap
Step 1: Pull a location you want to plot from Google maps.

map = suppressMessages(get_map(location = 'Europe', zoom = 4))

ggmap(map)

60
lat

50

40

Iris Malone Advanced Data Visualization in R November 6, 2015 65 / 68

ggmap

Step 2: Get geocoordinates for points or locations you’re interested in.

europegps = suppressMessages(geocode(c(
"Lisbon, Portugal",
"Eiffel Tower",
"Berlin, Germany",
"Crimea, Ukraine"), source="google"))
europegps

## lon lat
## 1 -9.139337 38.72225
## 2 2.294481 48.85837
## 3 13.404954 52.52001
## 4 34.102417 44.95212

Iris Malone Advanced Data Visualization in R November 6, 2015 66 / 68

ggmap
Step 3: Add geocoordinates to ggmap! It’s just like working with another
ggplot object.

ggmap(map) + geom_point(aes(x=europegps$lon, y = europegps$lat

lwd = 4, colour = "red") + ggtitle("Place I would like to

Place I would like to Visit!

60
lat

50

40
Iris Malone Advanced Data Visualization in R November 6, 2015 67 / 68
Summary

ggplot is super powerful

Iris Malone Advanced Data Visualization in R November 6, 2015 68 / 68

Summary

ggplot is super powerful but kind of annoying to learn

Iris Malone Advanced Data Visualization in R November 6, 2015 68 / 68

Summary

ggplot is super powerful but kind of annoying to learn

Ability to make complicated graphs awesome

Iris Malone Advanced Data Visualization in R November 6, 2015 68 / 68

Summary

ggplot is super powerful but kind of annoying to learn

Ability to make complicated graphs awesome
Benefits outweigh the start-up costs

Iris Malone Advanced Data Visualization in R November 6, 2015 68 / 68

Summary

ggplot is super powerful but kind of annoying to learn

Ability to make complicated graphs awesome
Benefits outweigh the start-up costs

Iris Malone Advanced Data Visualization in R November 6, 2015 68 / 68