Stat 437 Lecture Notes 2a

Xiongzhi Chen

Washington State University

Advanced visualization via ggplot2

Overview

We will cover: (a) faceting; (b) annotating a plot; (c) manually setting some scales; (d) adjusting some guides (e.g., legends and axis); (e) mathematical expressions in plots.

The contents for (a) to (d) are based on Chapter 1 of the book “R for data science” by Wickham and Grolemun, and on Chapters 5, 6 and 7 of the book “ggplot2: elegant graphs for data analysis” by Hadley Wickham.

Faceting

The mpg dataset

mpg1 is a subset of mpg; class refers to “type” of car, and drv type of drive

> # show levels of a factor
> levels(mpg1$class)
[1] "2seater"    "compact"    "midsize"    "minivan"   
[5] "pickup"     "subcompact" "suv"       
> levels(mpg1$drv)
[1] "4" "f" "r"
  • levels of a factor are ordered numerically and then alphabetically; this explains why “4” is ordered before “f” and “r”, and why “2seater” before “compact”

Faceting

  • A data set can be split into subsets based on some criteria. For example, the mileage dataset mpg can be split into 7 subsets according the 7 levels of class, or into 3 subsets according to the 3 levels of drv.

  • We have already discussed using aesthetics (e.g., color and shape) to compare or visualize subsets. Faceting takes an alternative approach by creating the same graph for each subset/subgroup of a data set and displaying the graphs in a table.

Two types of faceting

Two types of faceting are provided by ggplot2:

  • facet_grid produces a 2d grid of panels defined by variables which form the rows and columns
  • facet_wrap produces a 1d ribbon of panels that is wrapped into 2d

Two types of faceting

Difference between facet_grid and facet_wrap:

Faceting: facet_wrap

Basic syntax:

facet_wrap(facets, nrow = NULL, ncol = NULL,
  scales = "fixed", labeller = "label_value")

  • The facets can be specified by ~variable or vars(variable)

  • nrow (or ncol) sets number of rows (or columns) in which graphs are displayed

  • We will talk about scales and labeller later

  • Use ?facet_wrap to get more information

mpg data set

A quick look at factors drv, cyl, class and manufacturer

> library(ggplot2)
> # obtain levels of `drv`
> levels(mpg$drv)
[1] "4" "f" "r"
> 
> unique(mpg$cyl)
[1] 4 6 8 5
> 
> levels(mpg$class)
[1] "2seater"    "compact"    "midsize"    "minivan"   
[5] "pickup"     "subcompact" "suv"       
> 
> levels(mpg$manufacturer)
 [1] "audi"       "chevrolet"  "dodge"      "ford"      
 [5] "honda"      "hyundai"    "jeep"       "land rover"
 [9] "lincoln"    "mercury"    "nissan"     "pontiac"   
[13] "subaru"     "toyota"     "volkswagen"

Note how levels of a factor are ordered.

facet_wrap: illustration

Plot displ versus hwy for each level of class via faceting with rowwise layout:

> # build a base layer
> p1= ggplot(data = mpg)+
+   geom_point(mapping = aes(x = displ, y = hwy))
> 
> # add faceting via `class` to p1
> p2 = p1 + facet_wrap(~class, nrow = 2)
  • note the use of ~class; class has 7 levels
  • nrow specifies in how many rows the plots should be displayed

facet_wrap: illustration

> p2

facet_wrap: illustration

> p1 + facet_wrap(vars(class), nrow = 2) 

> # Same as: p1 + facet_wrap(~class, nrow = 2)

facet_wrap: illustration

Plot displ versus hwy for each level of class via faceting with columnwise layout:

> p1= ggplot(data = mpg)+
+   geom_point(mapping = aes(x = displ, y = hwy))
> 
> p2a = p1 + facet_wrap(vars(class), ncol = 3)
  • note the use of var(class), equivalent to ~class; class has 7 levels
  • ncol specifies in how many columns the plots should be displayed

facet_wrap: illustration

> p2a

facet_wrap: illustration

> # Plot `displ` versus `hwy`, faceting via `cyl` and `drv`
> ggplot(mpg,aes(x=displ,y=hwy))+geom_point()+
+ facet_wrap(cyl~drv)

Faceting: facet_grid

facet_grid forms a matrix of panels defined by row and column faceting variables. Basic syntax:

facet_grid(rows = NULL, cols = NULL, 
  scales = "fixed",labeller = "label_value")

  • rows (or cols): variables that define faceting groups on the row (or column) dimension

  • rows = NULL, cols = NULL takes the form of variable1 ~ variable2, .~ variable2 or variable1 ~.

  • Use ?facet_grid to obtain details

Facet grid via column variable

Plot displ versus hwy, faceting with drv columnwise:

> p1+ facet_grid(. ~ drv)

Facet grid via row variable

Plot displ versus hwy, faceting with drv rowwise:

> p1+ facet_grid(drv ~ .)

Facet grid with two variables

Plot displ versus hwy, faceting with drv and cyl:

> p1+ facet_grid(drv ~ cyl)

Visualization with >=3 factors

Recap on mpg data set:

# A tibble: 6 x 6
  manufacturer displ class   drv     cyl   hwy
  <fct>        <dbl> <fct>   <fct> <int> <int>
1 audi           1.8 compact f         4    29
2 audi           1.8 compact f         4    29
3 audi           2   compact f         4    31
4 audi           2   compact f         4    30
5 audi           2.8 compact f         6    26
6 audi           2.8 compact f         6    26

Visualization with >=3 factors

Plot displ versus hwy, faceting with drv and cyl, coloring points by class, assigning shapes by manufacturer ( chevrolet or toyota):

> mpg1 = mpg %>% 
+   filter(manufacturer %in% c("chevrolet","toyota"))
> 
> p1c = ggplot(mpg1, aes(x = displ, y = hwy))+theme_bw()+
+   geom_point(aes(colour=class,shape=manufacturer))+
+   facet_grid(drv~cyl)

Note: in mpg1, a combination of distinct levels or values of class, manufacturer, drv, and cyl may not have any observation

Visualization with >=3 factors

> p1c

Annotating a plot: Part I

Two commands for annotation

Sometimes one would like to annotate a plot to make it more informative. There are two commands for this purpose:

  • geom_text() adds text directly to a plot
  • geom_label() draws a rectangle around text

Syntax for commands

Basic syntax:

geom_text(mapping = NULL, data = NULL, 
  parse=FALSE, inherit.aes = TRUE)

geom_label(mapping = NULL, data = NULL,
  parse=FALSE, inherit.aes = TRUE)
  • mapping: set of aesthetic mappings created by aes(); if specified and inherit.aes = TRUE (the default), mapping is combined with the default mapping at the top level of the plot
  • mapping must be supplied if there is no plot mapping

Syntax for commands

Basic syntax (continued):

  • data: data to be displayed in this layer; if NULL, the default, data are inherited from the plot data as specified in the call to ggplot()

  • parse: if TRUE, labels will (often) be parsed into (math) expressions; FALSE by default

Use ?geom_text and geom_label to get more information

Recap on mpg data set

> head(mpg %>% select(class,drv,cyl,hwy))
# A tibble: 6 x 4
  class   drv     cyl   hwy
  <fct>   <fct> <int> <int>
1 compact f         4    29
2 compact f         4    29
3 compact f         4    31
4 compact f         4    30
5 compact f         6    26
6 compact f         6    26
> levels(mpg$drv)
[1] "4" "f" "r"
> unique(mpg$cyl)
[1] 4 6 8 5
> levels(mpg$class)
[1] "2seater"    "compact"    "midsize"    "minivan"   
[5] "pickup"     "subcompact" "suv"

geom_label: illustration

Plot displ vs hwy and annotate each pair of observations from “(displ,hwy)” by cyl type

> # build base layer
> p = ggplot(mpg,aes(displ,hwy))
> 
> # add label via `cyl`; "label" is an aesthetic
> p1 = p +geom_label(aes(label=cyl),nudge_x = 0.05)
> 
> # `nudge_x` horizontally adjusts labels to offset
> #  text from points
  • note the use of aes(label=cyl) in geom_label, aes(displ,hwy) in ggplot, and non-existence of geom_... for ggplot
  • nudge_y: vertically adjusts labels

geom_label: illustration

> p1

geom_label: illustration

> # Color `label` via `class`
> p+geom_label(aes(label=cyl,color = class))

geom_text: illustration

Plot displ vs hwy and annotate each pair of observations from “(displ,hwy)” by cyl type

> q = ggplot(mpg,aes(displ,hwy))+geom_text(aes(label=cyl))
  • note the use of aes(label=cyl) in geom_text, aes(displ,hwy) in ggplot, and non-existence of geom_... for ggplot

  • compare the above command with

    ggplot(mpg,aes(displ,hwy))+geom_label(aes(label=cyl))

geom_text: illustration

> q

geom_text: illustration

> # Color `label` by `class`
> ggplot(mpg,aes(displ,hwy))+
+   geom_text(aes(label=cyl,colour = class))

Annotating a plot: Part II

The command annotate

annotate adds geometric objects to a plot. But unlike a typical geom function, properties of the geometric objects are not mapped from variables of a data frame. Instead, they are passed in as vectors.

This is useful for adding small annotations (such as text labels), or if data in vectors instead of data frame need to be added.

The command annotate

Basic syntax:

annotate(geom, x = NULL, y = NULL,xmin = NULL, xmax = NULL,
  ymin = NULL, ymax = NULL)

  • geom: name of object to use for annotation, such as “text” or “segment”

  • x, y, xmin, ymin, xmax, ymax: positioning aesthetics, at least one of which must be specified

Use ?annotate to get more information on annotate

The command annotate

  • For annotating using text, use the following:

    annotate("text",x=NULL,y=NULL,label=NULL)

    for which the lengths of x, y and label must be compatible

  • For annotating using a rectangle, use the following:

    annotate("rect", xmin = NULL, xmax = NULL, 
   ymin = NULL, ymax = NULL)

  • There are other choices as annotations, use ?annotation to get more information

annotate: illustration

Plot displ vs hwy and add annotation “Some text”:

> # create base layer
> p = ggplot(mpg, aes(x = displ, y = hwy)) + geom_point()
> 
> 
> # add annotation "Some text" at positions 
> #(2,25), (3,25), (4,25) and (5,25)
> p1 = p + annotate("text", x = 2:5, y = 25,
+                   label = "Some text")
  • note the existence of geom_point() for ggplot
  • x = 2:5, y = 25 give labels positions (2,25), (3,25), (4,25) and (5,25) in coordinates

annotate: illustration

> p1

annotate: illustration

Plot displ vs hwy and add rectangle as annotation:

> # add a rectangle via "rect" whose size is determined
> # by `xmin`, `xmax`, `ymin` and `ymax`
> 
> p2 = p + annotate("rect", xmin = 3, xmax = 4.2, 
+         ymin = 12, ymax = 21, alpha = .2)
> 
> # `alpha` controls the degree of transparency of 
> # annotation; it can also control the transparency 
> # of ggplot2 geometric objects

annotate: illustration

> p2

annotate: illustration

Plot displ vs hwy and add a math expression as annotation:

> p2 = p + annotate("text", x = 4, y = 40, 
+                   label = "italic(R) ^ 2 == 0.75",
+                   parse = TRUE)
> 
> # `parse` converts the string "italic(R) ^ 2 == 0.75" into
> # a mathematical expression using latex syntax;
  • “italic(R) ^ 2 == 0.75” is equivalent to \(R^{2}=0.75\)
  • italic(R) sets the font of “R” to be italic; ^2 sets 2 as superscript
  • x = 4, y = 40 sets the position of annotation at coordinates (4,40)

annotate: illustration

> p2

License and session Information

License

> sessionInfo()
R version 3.5.0 (2018-04-23)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.1252 
[2] LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods  
[7] base     

other attached packages:
[1] ggplot2_3.1.0 knitr_1.21   

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.0       rstudioapi_0.8   bindr_0.1.1     
 [4] magrittr_1.5     tidyselect_0.2.5 munsell_0.5.0   
 [7] colorspace_1.3-2 R6_2.3.0         rlang_0.3.0.1   
[10] dplyr_0.7.8      stringr_1.3.1    plyr_1.8.4      
[13] tools_3.5.0      revealjs_0.9     grid_3.5.0      
[16] gtable_0.2.0     xfun_0.4         withr_2.1.2     
[19] htmltools_0.3.6  assertthat_0.2.0 yaml_2.2.0      
[22] lazyeval_0.2.1   digest_0.6.18    tibble_1.4.2    
[25] crayon_1.3.4     bindrcpp_0.2.2   purrr_0.2.5     
[28] glue_1.3.0       evaluate_0.12    rmarkdown_1.11  
[31] labeling_0.3     stringi_1.2.4    compiler_3.5.0  
[34] pillar_1.3.1     scales_1.0.0     pkgconfig_2.0.2