5/12/2017

ggplot2


install.packages(c('ggplot2','data.table')
library(ggplot2) # plotting
library(data.table) #fread
install.packages('tidyverse')
library(tidyverse) #includes ggplot2, dplyr, tidy, + more



Grammar of ggplot

Geoms

Aesthetics

  • Assign coordinates (x,y)
  • color, fill, shape, size, alpha ++
  • aes() maps data to the geom

ggplot()+geom_point(data, aes(x, y))

Dataset

Southern California Burritos

url<-'https://raw.githubusercontent.com/collnell/burritos/master/sd_burritos.csv'
ritos <- fread(url)

Building a plot

ggplot()

Mapping data & aes

ggplot()+
  geom_histogram(data=ritos, aes(x=Cost))

Aesthetics

Data & aes() mapping can be applied to each geom:

ggplot()+
  geom_histogram(data=ritos, aes(x=Cost, fill=Chips), binwidth = .25)

Or to all layers:

ggplot(data=ritos, aes(x=Cost, fill=Chips))+
  geom_histogram()

Or between both:

ggplot(data=ritos, aes(x=Cost))+
  geom_histogram(aes(fill=Chips))

Layering geoms

Multiple geoms can be plotted together using '+'

Use geom_boxplot to examine burrito cost by recommendation (Rec):

ggplot(data=ritos, aes(x=Rec, y=Cost))+
  geom_boxplot()

Layering geoms

Add layer showing data points with boxplot:

ggplot(data=ritos, aes(x=Rec, y=Cost))+
  geom_boxplot()+
  geom_point()

Mapping style aesthetics

Color, fill, shape, size, and alpha can be mapped to variables inside 'aes()'.

Create a scatterplot with Volume on the x-axis and Cost on y-axis

Start with base plot:

k<-ggplot(data=ritos, aes(x=Volume, y=Cost))
k

Add points for scatterplot:

k+geom_point()

Map levels of a variable using shapes

ritos$Rec<-as.factor(ritos$Rec)
levels(ritos$Rec) #aes maps to levels of a variable
[1] "no"  "yes"
k+geom_point(aes(shape=Rec))

Map geom color to show whether or not the burrito had guacaomle (Guac)

k+geom_point(aes(shape=Rec, color=Guac))

Size

k+geom_point(aes(shape=Rec, size=Taste))

##
Alpha = transparency

k+geom_point(aes(shape=Rec, alpha=Taste))

Setting vs. mapping aesthetics

You can also set aesthetic properties manually by assigning them outside of 'aes()'

k + geom_point(color='blue')

k + geom_point(aes(color=Taste), shape=21, alpha=.75, size=3)

#alpha ranges from 0 (transparent) to 1 (opaque)
# size is in mm

Fill & color assign color to different elements of the geom.

k + geom_point(aes(color=Taste), alpha=.5, size=3)

#alpha ranges from 0 (transparent) to 1 (opaque)
# size is in mm

Shapes:

k + geom_point(aes(fill=Taste), color='red',shape=21, alpha=.5, size=3)

##

k + geom_point(aes(fill=Taste), color='red',shape=21, alpha=.5, size=3, stroke = 3)

geom_smooth

Add trend line:

k+geom_point()+
  geom_smooth()

## geom_smooth
method - lm, glm, loess

?methods

k+geom_point()+
  geom_smooth()

geom_smooth

k+geom_point()+
  geom_smooth(method='lm', se=F)

Scales

Control the mapping of data and aesthetics
scale_x_reverse()
scale_y_reverse()
scale_x_log10()

ggplot(ritos, aes(log10(Volume), Cost))+
  geom_point()+
  geom_smooth(method='lm', se=F)

Color scales

Color scales

#install.packages('RColorBrewer')
library(RColorBrewer)

display.brewer.all()

ggplot(ritos, aes(Volume, Cost))+
  geom_point(aes(color = Taste), size=2)+
  geom_smooth(method='lm', se=F, color='black', lty='dashed')+
  scale_x_log10()+
  scale_color_gradient(low='red', high='blue')

Scales

scale_color_gradient - sequential color scale
scale_color_gradient2 - diverging color scale
scale_fill_gradient
scale_shape_discrete
scale_shape_manual - supply own values

scale_ + color/fill/size/shape/linetype/alpha_ + gradient/discrete/manual

ggplot(ritos, aes(Volume, Cost))+
  geom_point(aes(shape = Rec), size=2)+
  geom_smooth(method='lm', se=F, color='black', lty='dashed')+
  scale_x_log10()+
  scale_shape_manual(values = c(3,6), labels = c('Throw it away','Eat it again'))

Coordinates

Occur after statistics, affect geom appearance

To make a pie chart:

?coord_polar
d<-last_plot()
d+coord_flip()

Plot and axis titles

d+labs(title='Title goes here', y='Volume', 
       x='Burrito prices', caption='caption appears here ')

## Multiple plots

install.packages('cowplot')
library(cowplot)

?plot_grid

Theme

g<-ggplot(ritos, aes(Volume, Cost))+
  geom_point(aes(color = Taste), size=2)+
  geom_smooth(method='lm', se=F, color='black', lty='dashed')+
  scale_color_gradient(low='red', high='blue')

g_bw<-g+theme_bw()

g_min<-g+theme_minimal()

plot_grid(g_bw, g_min, nrow=1, ncol=2, labels=c('theme_bw','theme_minimal'))

Themes

Or independent elements can be set manually

g_bw + theme(legend.position = 'none', plot.background = element_rect(fill='green'),axis.text.y=element_text(angle=75, size=20))

g_bw+theme_mooney()

Resources