Introduction to R Graphics
Le Yan
HPC@LSU
h2
Disclaimer
R Graphic Systems
Outline
DataSets
First Plot in R
First Plot in R
We can use the plot() function in the base plot system to create a scatter plot:
# Simply specify the x and y variables.
plot(pressure$temperature,pressure$pressure)
Since there are only two variables, we can simply run:
plot(pressure)
More Plot Types
- The
typeargument ofplot()can be used to specify plot type
Line plot with “l”:
plot(pressure,type="l")
Or dot and line with “b”:
plot(pressure,type="b")
Adding More Layers
# Create the plot with title and axis labels.
plot(pressure,type="l",
main="Vapor Pressure of Mercury",
xlab="Temperature",
ylab="Vapor Pressure")
# Add points
points(pressure,size=4,col='red')
# Add annotation
text(150,700,"Source: Weast, R. C., ed. (1973) Handbook \n
of Chemistry and Physics. CRC Press.")
Boxplot
Boxplot
boxplot(hwy ~ cyl, data=mpg)
# Use the title function to add title and labels.
title("Highway Mileage per Gallon",
xlab = "Number of cylinders",
ylab = "Mileage (per gallon)")
Histogram
- Function
hist()can used to create histograms
hist(mpg$hwy)
hist(mpg$hwy, breaks=c(5,15,25,30,50))
Curve
The curve() function draws a function over a specified range.
curve(cos,-3*pi, 3*pi)
title("Cosine Function")
# The abline() function adds one or more straight lines to the current plot
abline(h=c(-1,0,1),
col = 2, lty = 2, lwd = 1.5)
Panel Grid of Plots
plot(airquality)
Saving Plots to Files
Example: saving to a PNG file
png("test.png",width=5*240,height=3*240)
plot(pressure, type="l")
points(pressure,col="red")
dev.off()
Here is the saved graph:
ggplot2 Package
qplot Function
Scatterplot with ```qplot``` Function
- Data represented by points
qplot(weightLb, heightIn, data=heightweight, geom="point")
- Data represented by labels
qplot(weightLb, heightIn, data=heightweight, geom ="text", label=ageYear)
A Fancier Plot
A Fancier Plot
Basic Concepts of ggplot2
ggplot(heightweight, # What data to use
aes(x=weightLb,y=heightIn)) + # Aesthetic specifies variables
geom_point() # Geom specifies visual marks
This is equivalent to:
qplot(weightLb, heightIn, data=heightweight, geom="point")
Histogram with ggplot2
ggplot(mpg,aes(x=hwy)) + geom_histogram(binwidth=5, fill="white", color="black")
Contour Plots with ggplot2
ggplot(faithfuld, aes(waiting, eruptions, z = density))+
geom_raster(aes(fill = density)) +
geom_contour(colour = "white")
Maps with ggplot2
- Combined with the
mapspackage, one can create geographical graphs
east_asia <- map_data("world",
region=c("Japan","China","North Korea","South Korea"))
ggplot(east_asia, aes(x=long,y=lat,group=group, fill=region)) +
geom_polygon(color="black") +
scale_fill_brewer(palette="Set2")
List of Geoms in ggplot2
Customizing Appearance of Data Points
ggplot(heightweight, aes(x=weightLb,y=heightIn)) +
geom_point(shape=13,size=5,color='red',alpha=0.5)
List of Symbols
- There are 25 built-in shapes
Notes on Colors
Adding More Layers to A Plot
- New layers can be added to a plot by using
geom_xxxfunctions
ggplot(heightweight, aes(x=weightLb,y=heightIn)) +
geom_point() +
geom_quantile(quantiles = c(0.25,0.5,0.75)) +
geom_text(label=rownames(heightweight), vjust=-0.5)
More on Aesthetic Mapping
Mapping Discrete Variables to Aesthetic Properties
- Discret data values can be mapped to an aesthetic value to group data points
Example: use the “sex” variable to group data points by shape and color:
ggplot(heightweight, aes(x=weightLb,y=heightIn)) +
geom_point(aes(shape=sex,color=sex))
Specify the shapes and colors manually (more on stat function later):
ggplot(heightweight, aes(x=weightLb,y=heightIn)) +
geom_point(aes(shape=sex,color=sex),size=4) +
scale_shape_manual(values=c(1,4)) +
scale_color_manual(values=c("blue","green"))
Mapping Continuous variables to Aesthetic Properties
- Continuous variables can be mapped to aesthetic values too
ggplot(heightweight, aes(x=weightLb,y=heightIn)) +
geom_point(aes(shape=sex,color=sex,size=ageYear))
Adding Fitted Models
Use stat_smooth function to add a fitted model to the plot:
ggplot(heightweight, aes(x=weightLb,y=heightIn)) +
geom_point(aes(shape=sex,color=sex),size=4) +
scale_shape_manual(values=c(1,4)) +
scale_color_manual(values=c("blue","green")) +
stat_smooth(method = lm, level=0.95)
Moving the color=sex statement to the ggplot function produces two lines:
ggplot(heightweight, aes(x=weightLb,y=heightIn, color=sex)) +
geom_point(aes(shape=sex),size=4) +
scale_shape_manual(values=c(1,4)) +
scale_color_manual(values=c("blue","green")) +
stat_smooth(method = lm, level=0.95)
Labeling individual points
To label data points, use either annotate or geom_text
ggplot(heightweight, aes(x=weightLb,y=heightIn)) +
geom_point(aes(shape=sex,color=sex,size=ageYear)) +
annotate("text",x=150,y=68,label="Some label",color="darkgreen",size=12)
ggplot(heightweight, aes(x=weightLb,y=heightIn)) +
geom_point(aes(shape=sex,color=sex,size=ageYear)) +
geom_text(aes(label=ageYear),vjust=0.5)
Stat Functions
- Some plots visualize a transformation of the original dataset.
- Use a
stat_xxxfunction to choose a common transformation to visualize.
We have seen the stat_smooth() function:
ggplot(heightweight, aes(x=weightLb,y=heightIn, color=sex)) +
geom_point(aes(shape=sex),size=4) +
scale_shape_manual(values=c(1,4)) +
scale_color_manual(values=c("blue","green")) +
stat_smooth(method = lm, level=0.95)
Stat Functions
Another example: stat_bin() function creates a frequency count:
ggplot(mpg,aes(x=hwy)) + stat_bin(binwidth = 5)
Stat Functions
Density plot with stat_density:
ggplot(mpg,aes(x=hwy)) + stat_density()
Or the same plot with geom_histogram:
ggplot(mpg,aes(x=hwy)) + geom_histogram(stat="density")
Saving Plot to An Object
- A
ggplotplot can be saved in an object - More convenient when you are experienting
p <- ggplot(heightweight, aes(x=weightLb,y=heightIn))
p + geom_point(aes(shape=sex,color=sex,size=ageYear))
# Here we use the saved plot object "p"
p + geom_smooth(method=lm)
Saving Plots to Files
Plot Titles
To add a title, use either ggtitle or labs(title=)
p <- ggplot(heightweight, aes(x=weightLb,y=heightIn, color=sex)) +
geom_point(aes(shape=sex),size=4) +
scale_shape_manual(values=c(1,4)) +
scale_color_manual(values=c("blue","green"))
p + ggtitle("Height ~ weight of school children")
Note the title is left-aligned by default.
Axis Labels
To add axis labels, use either (x|y)lab or labs(x=,y=)
p + ggtitle("Height ~ weight of school children") +
xlab("Weight (lbs)") + ylab("Height (inch)")
Legend Titles
- Use
labs(<aes>=)to specify legend titles
p + ggtitle("Height ~ weight of school children") +
xlab("Weight (lbs)") + ylab("Height (inch)") +
labs(color='Gender', shape='Gender')
Legends
- Use the
guidesfunction to set legend type for each aesthetic properties.
Before:
p <- ggplot(heightweight, aes(x=weightLb,y=heightIn,color=ageYear)) +
geom_point(aes(shape=sex))
p
After:
p + guides(shape='none',color='legend')
Themes
- Themes decide the appearance of a plot
ggplot2provides a few pre-defined themes for users to choose from
The classic theme:
p <- ggplot(heightweight, aes(x=weightLb,y=heightIn, color=sex)) +
geom_point(aes(shape=sex),size=4)
p + theme_classic()
The dark theme:
p + theme_dark()
Package ggthemes
- Additional themes are available from the
ggthemespackage
Example: Excel theme
p + theme_excel()
Fine-tuning the Theme
Removing the grid lines:
p + theme_bw() +
theme(panel.grid = element_blank())
Fine-tuning the Theme
Or just removing the vertical ones:
p + theme_bw() +
theme(panel.grid.major.x = element_blank(),
panel.grid.minor.x = element_blank())
Customizing Fonts
Change the base size and font family:
p + theme_bw(base_size = 24, base_family = "Times")
Or fine tune each element:
p + theme_bw(base_size = 24, base_family = "Times") +
theme(legend.title = element_text(size=20,color="blue"),# Legend title
legend.text = element_text(size=18,color="red"), # Legend text
axis.title.x = element_text(size=18,color="red"), # X axis label
axis.title.y = element_blank(), # Remove Y axis label
)
The element_blank() function can be used to remove undesired elements.
Changing Legend Position
p + theme_bw(base_size = 24, base_family = "Times") +
theme(legend.position = "bottom")
p + theme_bw(base_size = 24, base_family = "Times") +
theme(legend.position = c(0.9,0.1))
List of Theme Elements
Reset the default theme
- The default them is
theme_grey() - Use
theme_set()to change the default
With old default:
p
With new default:
theme_set(theme_light())
p
Creating Your Own Theme
- You can create your own theme and reuse later:
mytheme <- theme_bw(base_size = 24, base_family = "Times") +
theme(legend.title = element_text(size=20,color="blue"),# Legend title
legend.text = element_text(size=18,color="red"), # Legend text
axis.title.x = element_text(size=18,color="red"), # X axis label
axis.title.y = element_blank(), # Remove Y axis label
)
p + mytheme
Coordination systems
Coordination systems
Original:
g <- ggplot(mpg,aes(x=hwy)) + geom_histogram(binwidth=5, fill="white", color="black")
g
With flipped coorinates:
g + coord_flip()
Coordination systems
Original:
g
With transformed Y coordinate:
g + coord_trans(y="sqrt")
Axis Limits
- Use the
xlim()andylim()functions to set the range of axes:
p + theme_light() +
xlim(0,200) +
ylim(50,100)
Scales
X and Y scales
scale_x_continuous: scale for X, which is a continuous variable
p + theme_bw() +
ylim(50,100) +
scale_x_continuous(limits=c(0,200),
breaks=c(50,110,170),
labels=c("Thin","Medium\nSize","Chubby"))
X and Y scales
p + theme_economist_white() +
scale_x_log10(breaks=c(10,20,50,100,200),
limits=c(5,500)) + # Plot X on a log10 scale
scale_y_reverse() # Reverse the Y scale
Legend Labels
- Scale functions can be used to customize legend labels
- Color, shape, size, fill etc.
ggplot(mpg,aes(x=drv,y=cty,fill=drv)) +
geom_boxplot()
ggplot(mpg,aes(x=drv,y=cty,fill=drv)) +
geom_boxplot() +
scale_fill_discrete(limits=c("f","r","4"),
labels=c("Front","Rear","4 Wheel Drive"))
Other scales
By default:
ggplot(mpg,aes(x=displ,y=hwy,size=cyl,
color=drv,shape=fl)) +
geom_point(aes(alpha=cty))
Re-scaled
ggplot(mpg,aes(x=displ,y=hwy,size=cyl,color=drv, alpha=cty)) +
geom_point() +
scale_size_identity() + # Use the values of "cyl" variable for size
scale_color_manual(values=c("darkblue","rosybrown2","#24FA22")) +
scale_alpha_continuous(range=c(0.1,1))
Faceting
facet_grid: create a row of panels defined by the variable “drv”:
ggplot(mpg, aes(displ, hwy)) +
geom_point() +
facet_grid(. ~ drv)
Facet_grid
facet_grid: creates a column of panels defined by the variable “fl”:
ggplot(mpg, aes(displ, hwy)) +
geom_point() +
facet_grid(fl ~ .)
Facet_grid
facet_grid: creates a matrix of panels defined by the variables “fl” and “drv”:
ggplot(mpg, aes(displ, hwy)) +
geom_point() +
facet_grid(fl ~ drv)
Facet_wrap
facet_wrap: wraps 1d sequence of panels into 2d:
ggplot(mpg, aes(displ, hwy)) +
geom_point() +
facet_wrap(~class, nrow=3)
