14 Base R plots

Graphical presentations in Statistics

  • Graphs are helpful for presenting a summary of data and results of a statistical analysis

  • John W. Tukey, the father of exploratory data analysis once said

The greatest value of a picture is when it forces us to notice what we never expected to see.

Base R plot functions

Graphical presentations of data

  • There are several graphs available to use for describing data, and the selection of the most appropriate graph depends on the data type and the research objectives

  • Quantitative data

    • Histogram

    • Boxplot

    • Scatter plot

  • Qualitative data

    • Bar chart

    • Pie chart

  • Bivariate analysis involves two variables, depending on the combinations of the variables, i.e., qualitative or quantitative, there are different ways of presenting data graphically

  • Quantitative-qualitative combination

    • Histogram and boxplot can be used for different levels of a qualitative variable

  • Quantitative-quantitative and qualitative-qualitative combinations

    • Bar chart and scatter plot can be used

1 Histogram

  • The function hist() is used to obtain a histogram of a quantitative variable

  • Syntax of hist()

hist(x, ...)
  • x is a quantitative vector
hist(x = penguins$body_mass_g)

  • Some useful arguments of hist():

    • xlab, main, probability, etc.
hist(
  x = penguins$body_mass_g, 
  xlab = "Body mass", 
  main = "Histogram of penguins'
                  body mass" 
  )

hist(
  x = penguins$body_mass_g, 
  xlab = "Body mass",
  main = "Histogram of penguins'
                  body mass",
  breaks = 20, 
  probability = T 
  )

hist(
  x = penguins$body_mass_g, 
  xlab = "Body mass",
  main = "Histogram of penguins'
                  body mass",
  breaks = 20, 
  probability = T 
  )
#
lines(
  density(x = penguins$body_mass_g,
          na.rm = T),
  lwd = 3, col = "brown"
  )

Exercise 3.1.1

(use mtcars data frame to answer the followings)

  • Create a histogram of mpg with appropriate labels

  • Add density line to the plot obtained in Question 1.

2 Boxplot

  • Boxplot is a useful graphical tool that can be used to compare distribution of a quantitative variable at different levels of a qualitative variable

    • E.g. examine the distribution of body mass over different species of penguins

  • boxplot() function can be used for both univariate and bivariate analysis

    • boxplot(x) is used to obtain a boxplot of a single quantitative vector x

    • boxplot(formula, data) function is used for a bivariate analysis, where the formula species the quantitative and qualitative variables of interest

    • formula = quant_var ~ qual_var

    • data is a data frame that must contain quant_var and qual_var

boxplot(x = penguins$body_mass_g)

boxplot(formula = body_mass_g ~ species, 
        data = penguins)

Exercise 3.1.2

(use mtcars data frame to answer the followings)

  • Create a boxplot of qsec with appropriate labels.

  • Create a boxplot of mpg to compare its distribution at different levels of cyl

3 Scatter plot

  • The function plot(x, y) is used to obtain a scatter plot of two quantitative variables x and y

  • Some useful arguments of plot() function

    • xlab, ylab, main

    • pch (point type)

    • cex (size of points), etc.

#
plot(x = penguins$bill_length_mm, 
     y = penguins$flipper_length_mm)

plot(x = penguins$bill_length_mm, 
     y = penguins$flipper_length_mm,
     xlab = "Flipper length") 

plot(
  x = penguins$flipper_length_mm,
  y = penguins$bill_length_mm, 
  xlab = "Flipper length",
  pch = 20, 
  cex = 1.5, 
  col = "brown" 
  )

Scatter plot with a linear model fit

plot(
  x = penguins$flipper_length_mm,
  y = penguins$bill_length_mm, 
  xlab = "Flipper length",
  pch = 20,
  cex = 1.5, 
  col = "brown")
#
mod1 <- lm(bill_length_mm ~ flipper_length_mm, 
     data = penguins)
#
abline(mod1, col = "blue", lwd = 4)

  • lm() is for fitting a linear model

Exercise 3.1.3

(use mtcars data frame to answer the followings)

  • Create a scatter plot to examine the association between mpg and disp

  • Add the fit of a linear regression model mpg on disp to the plot obtained in Question 6

4 Bar chart

  • Bar chart is used to examine the distribution of a qualitative variable

  • The function barplot(height, ...) is used to obtain a bar chart in R, where height represents a frequency

  • table() function takes a qualitative variable as an argument and returns height, the frequency corresponding to each level of the qualitative variable

Frequency distribution of species 

table(penguins$species)
#> 
#>    Adelie Chinstrap    Gentoo 
#>       152        68       124
barplot(height = table(penguins$species))

Exercise 3.1.4

(use mtcars data frame to answer the followings)

  • Create a barchart if cyl

Bivariate analysis

  • The function par() has many arguments that can be used to produce high-quality graphs using base R plot functions

  • mfrow argument of par() is used to split a figure layout into a number of rows and columns

    • E.g. mfrow = c(2, 3) will split the figure layout into two rows and three columns

Distribution of bill_length_mm at different levels of species

par(mfrow = c(1, 3))
hist(x = penguins$bill_length_mm[penguins$species == "Adelie"])

par(mfrow = c(1, 3))
hist(x = penguins$bill_length_mm[penguins$species == "Adelie"])
hist(x = penguins$bill_length_mm[penguins$species == "Gentoo"])

par(mfrow = c(1, 3))
hist(x = penguins$bill_length_mm[penguins$species == "Adelie"])
hist(x = penguins$bill_length_mm[penguins$species == "Gentoo"])
hist(x = penguins$bill_length_mm[penguins$species == "Chinstrap"])

Exercise 3.1.5

  • Association between bill and flipper lengths by species