ggplot and related plotting mechanisms

Oh, they have already created a book: https://ggplot2-book.org/index.html

Basic plots

Basic scatter plot setup

Two variables: x and y both continuous

library(ggplot2)
ggplot(df, aes(x=xcol, y=ycol))

where df is a dataframe that contains all information to make the ggplot. Plot will show up only after adding the geom layers shown later using + sign consequtively.

a1 <- ggplot(mtcars, aes(mpg, wt))
a1 + geom_point()

ggplot(mtcars, aes(mpg, wt)) +
  geom_point()

You can add color to the points inside the aes or aesthetic, either

ggplot(mtcars, aes(mpg, wt, color=cyl)) +
  geom_point()

or later,

a1 + geom_point(aes(color=cyl))

Bad example ahead, To have more access to the other parameters,

a3 <- ggplot(diamonds, aes(x=carat, y=price))
a3 + geom_point(aes(size=carat, shape=cut, color=color, stroke=carat))

Title, labels and themes

a1 + 
labs(title="Croods", subtitle="cars", x="miles per gallon", y="weight", caption = "(based on data from 1974 _Motor Trend_ US magazine)", tag="A")

add NULL if you want to remove something. (e.g., title=NULL) Some equivalent functions exists with xlab, ylab, ggtitle etc functions.

You can change other corresponding things too:

a1 + geom_point(aes(color=cyl)) + 
  labs(colour = "Cylinders")

Aesthetics

lines:

with color, these can be changed, size width of line in mm

linetype 0 = blank, 1 = solid, 2 = dashed, 3 = dotted, 4 = dotdash, 5 = longdash, 6 = twodash

More can be done with string containing 2, 4, 6, or 8 hexadecimal digits which give the lengths of consecutive lengths. For example, the string "33" specifies three units on followed by three off (i.e., dot dot dot blank blank blank) and "3313" (dddbbbdbbb)

lineend 'round', 'butt' (the default), or 'square'

linejoin 'round' (the default), 'mitre', or 'bevel'

Polygons

fill is the additional parameter here.

Point

shape

Themes

Jittered plot with another data:

a2 <- ggplot(mpg, aes(cty, hwy))
a2 + geom_jitter()

with some added color variation with other variables (there would be warning, but you get the idea):

a2 + geom_jitter(aes(color=manufacturer, shape=class))

Bar plots and Voilin plot

For these kind of plots, one variable should be as a factor (say the x variable), and the other should be a continuous/discrete variable.

ggplot(data, aes(x=name, y=value)) + 
  geom_bar(stat = "identity")

geom_bar takes color, fill, width as arguments as usual.

To make this circular, add

  + coord_polar(start = 0)

Often, the factors are sorted according to the factor's lavels. To correct those, you may try the reorder() function inside a with() call

# reorder is close to order, but is made to change the order of the factor levels.
mpg$class = with(mpg, reorder(class, hwy, median))

ggplot(mpg, aes(x=class, y=hwy, fill=class)) +
    geom_bar(stat='identity', alpha=0.5)

Note, if you don't give stat=identity, it will try to calculate the y values itself as the default is stat=count, eventually producing an error.

You can add error bars also using something like:

geom_errorbar( aes(x=name, ymin=value-sd, ymax=value+sd), width=0.4, colour="orange", alpha=0.9, size=1.3)

ROC plots

geom_roc()

Different coloring techniques

Add these lines to get different types of colors and shades

scale_fill_hue(c = 40)
scale_fill_brewer(palette = "Set1")
scale_fill_grey(start = 0.25, end = 0.75)
scale_fill_manual(values = c("red", "green", "blue") )

Different coloring proccess:

Grouping

I find that sometimes grouping gets extremely complex in many scenarios.

FilesExpand file tree

ggplot.md

Latest commit

History