In emeyers/SDS230: Tools for the class Data Exploration and Analysis

$\$

# makes sure you have all the packages we have used in class installed
#SDS230::update_installed_packages()


# install.packages("gapminder")

# install.packages("latex2exp")

library(latex2exp)

library(dplyr)


#options(scipen=999)


knitr::opts_chunk$set(echo = TRUE)

# hide all plot output - useful for printing the code
# knitr::opts_chunk$set(fig.show='hide')

set.seed(123)

$\$

Overview

• Plotting with ggplot

$\$

Part 1: Visualizations with ggplot

We can use the ggplot2 package, which is part of the tidyverse, to create much nicer looking graphics than using base R graphics. The ggplot2 library is modeled on Leland Wilkinson's "grammar of graphics" which creatse graphics from a combination of basic visual elements.

In the exercises below, we will learn how to use ggplot using the motor trends cars data set (mtcars) that comes with base R installation, and also the gapminder data.

A few resources to learn more about ggplot are:

• R for Data Science
• The ggplot cheatsheet

$\$

Part 1.1: Scatter plots

Let's create plots of the number of miles per gallon (mpg) cars get as a function of the weight of the car.

# install.packages("ggplot2")


library(ggplot2)


# base R
plot(mtcars$wt, mtcars$mpg)


# ggplot - global mapping
ggplot(data = mtcars, mapping = aes(x = wt, y = mpg)) +
    geom_point() 


# ggplot - shorter global mapping
ggplot(mtcars, aes(x = wt, y = mpg)) +
    geom_point() 


# ggplot - mapping in the geom
ggplot(data = mtcars) +   
    geom_point(mapping = aes(x = wt, y = mpg))

$\$

Part 1.2: Adding labels to plots

We can add labels to the plots using the labs() function. Arguments to labs() function include: - x: the label on the x-axis (you can also use the xlab() function) - y: the label on the y-axis (you can also use the ylab() function) - title: the title of the plot (you can also use the ggtitle() function) - subtitle: the title of the plot (you can also use the ggtitle() function)

If you just want to add x and y labels or a title, you can also use the xlab("label1"), ylab("label2") and/or the ggtitle() functions.

ggplot(mtcars, aes(x = wt, y = mpg)) +          
    geom_point()  +    
    labs(x = "Weight",
         y = "Miles per Gallon",
         title = "MPG as a function of weight", 
         subtitle = "ggplot is cool!")



# another way to add just x and y labels
ggplot(mtcars, aes(x = wt, y = mpg)) +          
    geom_point()  +    
    xlab("Weight") +   
    ylab("Miles per Gallon") + 
  ggtitle("MPG as a function of weight")

Remember, if you don't want exes label your axes

$\$

Part 1.3: Adding annotations to plots

We can add annotations to plots using the annotate("text", x = , y = , label = ) function.

ggplot(mtcars, aes(x = wt, y = mpg)) +          
    geom_point()  +    
    annotate("text", x = 4, y = 30, label = "AWESOME!")

$\$

Part 1.4: More aesthetic mappings

We can use other aesthetic mappings beyond position including: - color: different color scales are used for quantitative and categorical (factor) data - shape: should be categorical data - size: should be quantitative data

# add color based on the transmission type (is automatic or not)
ggplot(mtcars, aes(x = wt, y = mpg, color = am)) +
    geom_point()


# it is better to treat the transmission type as a categorical variable?
ggplot(mtcars, aes(x = wt, y = mpg, col = factor(am))) +  
  geom_point()


# can also try mapping transmission type to shape or size 
ggplot(mtcars, aes(x = wt, y = mpg, shape = factor(am))) +  
   geom_point()

ggplot(mtcars, aes(x = wt, y = mpg, size = am)) +  
   geom_point()

Question: When adding the variable automatic/manual transmission (am) to the scatter plot mpg vs. weight, do you think it is best to map am on to...?

a. color b. shape c. size

$\$

Part 1.5: Attributes vs. Aesthetics

Setting an aesthetic mapping maps a variable to a glyph property. This is done inside the aes() function.

Setting an attribute set a glyph property to a fixed value. This is done outside the aes() function.

# setting an aesthetic mapping
ggplot(mtcars) +
    geom_point(aes(x = wt, y = mpg, col = factor(am)))


# setting an attribute
ggplot(mtcars) + 
    geom_point(aes(x = wt, y = mpg), col = "red")

$\$

Part 1.6: Facets

Beyond comparing variables based on aesthetic mappings, you can compare categorical variables by splitting a plot into subplots, called facets, using facet_wrap()

# separate subplots for the two transmission types
ggplot(mtcars, aes(x = wt, y = mpg)) +      
  geom_point() +   
    facet_wrap(~am)


# One can also do facets in two dimensions
ggplot(mtcars, aes(x = wt, y = mpg)) +      
    geom_point() +   
    facet_wrap(am ~ cyl)

$\$

Part 1.7: Overplotting

Sometimes points overlap making it hard to estimate the number of points at a particular range of values.

We can control the transparency of points by changing their alpha values.

library(gapminder)


# a lot of overplotting
ggplot(gapminder, aes(x = gdpPercap, y = lifeExp)) +    
    geom_point()


# changing the transparency levels makes it a bit easier to see how many points are at a given x, y location
ggplot(gapminder, aes(x = gdpPercap, y = lifeExp)) +    
    geom_point(alpha = .1)

$\$

Part 1.8: Changing scales

Each visual attribute that has an aesthetic mapping has a default scales. We can change the scales used for each mapping using functions that start with scale_.

For example, we can change the x-scale from liner to logarithmic using scale_x_continuous(trans='log10'). Likewise we can change the color scale using scale_color_manual().

# changing the scale on the x-axis
ggplot(gapminder, aes(x = gdpPercap, y = lifeExp)) +    
    geom_point(alpha = .2) + 
  scale_x_continuous(trans='log10')


# mapping continents to colors, and adding my own color scale
ggplot(gapminder, aes(x = gdpPercap, y = lifeExp, col = continent)) +   
    geom_point(alpha = .2) + 
  scale_x_continuous(trans='log10') + 
  scale_color_manual(values = c("red", "yellow", "green", "blue", "purple"))

$\$

Part 1.9: geoms

We can use different geoms to create other types of plots.

# Let's create a plot that shows the GDP in the United States as a function of the year using the geom geom_line()
gapminder |> 
  filter(country == 'United States') |>     
  ggplot(aes(x = year, y = gdpPercap)) +      
    geom_line()


# Let's create a plot that shows the GDP in the United States as a function of the year using the geom geom_col()
gapminder |> 
  filter(country == 'United States') |>     
  ggplot(aes(x = year, y = gdpPercap)) +      
    geom_col()


# Let's MPG as a function of weight using the names of cars rather than just points for each car
mtcars |>
  tibble::rownames_to_column() |>
  ggplot(aes(x = wt, y = mpg)) + 
    geom_text(aes(label = rowname))


# Let's plot a histogram of the weights of cars
ggplot(mtcars, aes(x = wt)) +       
    geom_histogram(bins = 10)


# Let's create a boxplot of the weights of cars
ggplot(mtcars, aes(x = "", y = wt)) +       
    geom_boxplot()


# Let's create a side-by-side boxplot of the weights of cars depending on the number of cylinders the engine has
ggplot(mtcars, aes(x = factor(cyl), y = wt)) +      
  geom_boxplot()

$\$

Part 1.10: geoms continued

Violin and Joy plots are other ways to view distributions of data

# violin plot 
ggplot(mtcars, aes(x = factor(cyl), y = wt)) +  
    geom_violin()



library("ggridges")

# joy plot
ggplot(mtcars, aes(y = factor(cyl), x = wt)) +  
  geom_density_ridges()

Question: Can you figure out where the name "joy plot" comes from?

$\$

Part 1.11: geoms continued

We can also have multiple geom layers on a single graph by using the + symbol E.g ggplot(…) + geom_type1() + geom_type2()

# Create a scatter plot of miles per gallon as a function of weight and then add a smoothed line using geom_smooth() and a vertical lines using geom_vline()

ggplot(mtcars, aes(x = wt, y = mpg)) +      
    geom_point() +  
    geom_smooth() +
  geom_vline(xintercept = 3)

$\$

Part 1.12: Themes

We can also use different themes to change the appearance of our plot.

# Add theme_classic() to our plot

ggplot(mtcars, aes(x = wt, y = mpg)) +          
    geom_point()  +    
    xlab("Weight") +   
    ylab("Miles per Gallon") + 
    theme_classic() + 
  theme(                                              # modify the theme by:
        axis.text.y = element_blank(),                #  - turning off the y-axis text
        plot.background = element_rect(fill = "red")  # - making the background red
        )                                             # see ? theme for more options


# install.packages("ggthemes")
library(ggthemes)

ggplot(mtcars, aes(x = wt, y = mpg)) +          
    geom_point()  +    
    ggtitle("Cars!") + 
    theme_fivethirtyeight()

$\$

Part 1.13: Exeriment yourself

Try to create some interesting visualizations from either a data set we have used in the class, or a new data set you found.

$\$

emeyers/SDS230 documentation built on Jan. 18, 2024, 1:01 a.m.