In therbootcamp/BaselRBootcamp2017:

knitr::opts_chunk$set(comment=NA, fig.width=6, fig.height=6, echo = TRUE, eval = TRUE, message = FALSE, warning = FALSE, fig.align = 'center')

knitr::include_graphics("https://d21ii91i3y6o6h.cloudfront.net/gallery_images/from_proof/9296/small/1447173871/rstudio-hex-ggplot2-dot-psd.png")

Slides

Here are the introduction slides for this practical on plotting!

Overview

In this practical you'll practice plotting data with the ggplot2 package.

Cheatsheet

knitr::include_graphics("images/ggplot_cheatsheet_ss.png")

If you don't have it already, you can access the ggplot2 cheatsheet here https://www.rstudio.com/wp-content/uploads/2015/03/ggplot2-cheatsheet.pdf. This has a nice overview of all the major functions in ggplot2.

Examples

• The following examples will take you through the steps of creating both simple and complex plots with ggplot2. Try to go through each line of code and see how it works!

# -----------------------------------------------
# Examples of using ggplot2 on the mpg data
# ------------------------------------------------

library(tidyverse)         # Load tidyverse (which contains ggplot2!)

mpg # Look at the mpg data


# Just a blank space without any aesthetic mappings
ggplot(data = mpg)

# Now add a mapping where engine displacement (displ) and highway miles per gallon (hwy) are mapped to the x and y aesthetics
ggplot(data = mpg, 
       mapping = aes(x = displ, y = hwy))   # Map displ to x-axis and hwy to y-axis

#  Add points with geom_point()
ggplot(data = mpg, 
       mapping = aes(x = displ, y = hwy)) +
       geom_point()                         

# Again, but with some additional arguments

ggplot(data = mpg, 
       mapping = aes(x = displ, y = hwy)) +
       geom_point(col = "white",                # White borders
                  fill = "red",                 # Red filling
                  size = 3,                     # Larger size
                  shape = 21,                   # Point with border shape
                  alpha = .5,                   # Transparent points
                  position = "jitter") +        # Jitter the points         
         scale_x_continuous(limits = c(1, 10)) +  # Axis limits
         scale_y_continuous(limits = c(0, 50))


# Assign class to the fill aesthetic and add labels with labs()

ggplot(data = mpg, 
  mapping = aes(x = displ, y = hwy, fill = class)) +  # Change color based on class column
  geom_point(size = 3, 
                  position = 'jitter',
                  shape = 21, 
                  col = "white") +
  labs(x = "Engine Displacement in Liters",
       y = "Highway miles per gallon",
       title = "MPG data",
       subtitle = "Cars with higher engine displacement tend to have lower highway mpg",
       caption = "Source: mpg data in ggplot2")


# Add a regression line and use the black and white theme with theme_bw()

ggplot(data = mpg, 
       mapping = aes(x = displ, y = hwy)) +
  geom_point(size = 3, alpha = .9, mapping = aes(color = class)) + 
  geom_smooth(col = "blue", method = "lm") + 
  labs(x = "Engine Displacement in Liters",
       y = "Highway miles per gallon",
       title = "MPG data",
       subtitle = "Cars with higher engine displacement tend to have lower highway mpg",
       caption = "Source: mpg data in ggplot2") + 
  theme_bw()    # Use black and white theme


# Facet by class
ggplot(data = mpg) +
  geom_point(mapping = aes(x = displ, 
                           y = hwy, 
                           fill = factor(cyl)),
                           shape = 21,
                           col = "white", 
                           size = 2) +
  facet_wrap(~ class, nrow = 2, labeller = "label_both") + 
  scale_fill_brewer(palette = "Set1") + 
  labs(x = "Engine Displacement in Liters",
       y = "Highway miles per gallon",
       title = "MPG data",
       subtitle = "Cars with higher engine displacement tend to have lower highway mpg",
       caption = "Source: mpg data in ggplot2") + 
  theme_bw()        # Use the minimal theme


# Another fancier example

ggplot(data = mpg, 
       mapping = aes(x = cty, y = hwy)) + 
       geom_count(aes(color = manufacturer)) +     # Add count geom (see ?geom_count)
       geom_smooth(se = FALSE) +                   # smoothed line without confidence interval
       geom_text(data = filter(mpg, cty > 25), 
                 aes(x = cty,y = hwy, 
                     label = rownames(filter(mpg, cty > 25))),
                     position = position_nudge(y = -1), 
                                check_overlap = TRUE, 
                     size = 5) + 
       labs(x = "City miles per gallon", 
            y = "Highway miles per gallon",
            title = "City and Highway miles per gallon", 
            subtitle = "Numbers indicate cars with highway mpg > 25",
            caption = "Source: mpg data in ggplot2",
            color = "Manufacturer", 
            size = "Counts") +
        theme_bw()

Tasks

Getting the data and project setup

1. For this practical we'll play around with a few different datasets that are contained in different packages. The datasets, and the packages that contain them, are listed below. If you don't have any of these packages already, make sure to install them!

| Dataset| Package| |:------|:----| | ACTG175| speff2trial | | diamonds| ggplot2 | | Davis| car | | heartdisease | FFTrees |

1. Load the tidyverse, speff2trial, car, and FFTrees packages

library(tidyverse)
library(speff2trial)
library(car)
library(FFTrees)

Building a plot step-by-step

1. The diamonds dataset in the ggplot2 package shows information about 50,000 round cut diamonds. Print the diamonds dataset, it should look like this:

diamonds

Create the following blank plot

ggplot(data = diamonds, 
       mapping = aes(x = carat, y = price))

1. Now add points showing the relationship between the number of carats in the diamonds (carat) and its price (price)

ggplot(data = diamonds, 
       mapping = aes(x = carat, y = price)) + 
  geom_point()

1. Make the points transparent using the alpha argument to geom_point()

ggplot(data = diamonds, 
       mapping = aes(x = carat, y = price)) + 
  geom_point(alpha = .05)

1. Color the points by their cut

ggplot(data = diamonds, 
  mapping = aes(x = carat, y = price, color = cut)) + 
  geom_point(alpha = .05)

1. Create different plots for each value of cut using the facet_wrap function:

ggplot(data = diamonds, 
       mapping = aes(x = carat, y = price, color = cut)) + 
  geom_point(alpha = .05) + 
  facet_wrap(~ cut, nrow = 1)

1. Add a black, smoothed mean line to each plot using geom_smooth() (You can also try turning the line into a regression line using the method argument)

ggplot(data = diamonds, 
       mapping = aes(x = carat, y = price, color = cut)) + 
  geom_point(alpha = .05) + 
  facet_wrap(~ cut, nrow = 1) + 
  geom_smooth(color = "black")

Density geom with geom_density()

Create the following density plot of prices from the diamonds data using the following template:

• Set the data argument to diamonds
• Map carat to the x aesthetic
• Add a density geom with geom_density() and set the fill to "tomato1"
• Add labels
• Use the minimal theme with theme_minimal()

ggplot(data = XX, 
       mapping = aes(x = XX)) + 
       geom_density(fill = "XX") + 
       labs(x = "XX", 
            y = "XX", 
            title = "XX",
            subtitle = "XX",
            caption = "XX") +
       theme_minimal() 

ggplot(data = diamonds, 
       mapping = aes(x = carat)) + 
  geom_density(fill = "tomato1") + 
  labs(x = "Carats", 
       y = "Count", 
       title = "Diamond Prices",
       subtitle = "Created with ggplot!",
       caption = "Source: diamonds dataset") +
       theme_minimal()

Create the following densities of different facets by the diamond cut, and use different colors for each plot.

• Map log(price) (the logarithm of price) to the x aesthetic and cut to the fill aesthetic
• Create the density geom with geom_density() and add a black border color with the color argument
• Facet the plot by cut
• Add labels with labs(x, y, title, subtitle, caption)
• Using scale_fill_brewer(), change the color palette for the geom fillings to "YlOrRd"
• Use the black and white theme with theme_bw()
• Turn off the legend with theme(legend.position = "none")

ggplot(data = XX, 
       mapping = aes(x = log(XX), fill = XX)) + 
       geom_density(color = "XX") + 
       facet_wrap(~ XX, nrow = 1) + 
       labs(x = "XX", 
            y = "XX", 
            title = "XX",
            subtitle = "XX",
            caption = "XX") +
       scale_fill_brewer(palette = "XX") + 
       theme_bw() +
       theme(legend.position = "none")

ggplot(data = diamonds, 
       mapping = aes(x = log(price), fill = cut)) + 
       geom_density(color = "black") + 
       facet_wrap(~ cut, nrow = 1) + 
       labs(x = "Price (log-transformed)", 
            y = "Count", 
            title = "Diamond Prices",
            subtitle = "Separate plots indicate different cuts",
            caption = "Source: diamonds dataset from ggplot2") +
       scale_fill_brewer(palette = "YlOrRd") + 
       theme_bw() +
       theme(legend.position = "none")

Boxplot geom geom_boxplot()

Look at the help menu for geom_boxplot(). Then, create the following boxplot using the following template

ggplot(data = XX,
  mapping = aes(x = XX, y = log(XX), fill = XX)) + 
  geom_boxplot()  + 
  labs(y = "XX", 
       x = "XX", 
       color = "XX",
       title = "XX",
       subtitle = "XX") +
  scale_fill_brewer(palette = "XX") +
  theme_bw()

ggplot(data = diamonds,
  mapping = aes(x = cut, y = log(price), fill = cut)) + 
  geom_boxplot()  + 
  labs(y = "Price (log scale)", x = "Cut", color = "Cut", 
       title = "Distribution of diamond prices by cut", 
       subtitle = "Data come from a random sample of 1000 diamonds",
       caption = "Source: diamonds dataset from ggplot2") +
  scale_fill_brewer(palette = "YlGnBu") +
  theme_bw()

Demographic information of midwest counties in the US

Print the midwest dataset and look at the help menu to see what values it contains. It should look like this:

midwest

Using the following code as a template, create the Following plot showing the relationship between college education and poverty

ggplot(data = XX, 
    mapping = aes(x = XX, y = XX)) + 
    geom_point(aes(fill = XX, size = XX), shape = 21, color = "white") + 
    geom_smooth(aes(x = XX, y = XX)) +
    labs(
        x = "XX", 
        y = "XX", 
        title = "XX",
        subtitle = "XX",
        caption = "XX") + 
    scale_color_brewer(palette = "XX") + 
    scale_size(range = c(XX, XX)) +
    guides(size = guide_legend(override.aes = list(col = "black")), 
           fill = guide_legend(override.aes = list(size = 5))) +
    theme_bw()

ggplot(data = midwest, 
    mapping = aes(x = percollege, y = percpovertyknown)) + 
    geom_point(aes(fill = state, size = popdensity), shape = 21, color = "white") + 
    geom_smooth(aes(x = percollege, y = percpovertyknown)) +
    labs(
        x = "Percent with college education", 
        y = "Poverty rate", 
        title = "Midwest Data",
        subtitle = "States with higher college education rates tend to have lower poverty rates",
        caption = "Source: ggplot2 package") + 
    scale_color_brewer(palette = "Set1") + 
    scale_size(range = c(0, 12)) +
    guides(size = guide_legend(override.aes = list(col = "black")), 
           fill = guide_legend(override.aes = list(size = 5))) +
    theme_bw()

Create the following density plot showing the density of inhabitants with a college education in different states using the following template

ggplot(data = XX, 
       mapping = aes(XX, fill = XX)) + 
  geom_density(alpha = XX) + 
  labs(title = "XX", 
       subtitle = "XX",
       caption = "XX",
       x = "XX",
       y = "XX",
       fill = "XX") + 
  theme_bw()

ggplot(data = midwest, 
       mapping = aes(percollege, fill = state)) + 
  geom_density(alpha = 0.8) + 
  labs(title = "College education rates", 
       subtitle = "For 5 Midwest states",
       caption = "Source: midwest dataset in ggplot2",
       x = "Percent of inhabitants with a college education",
       y = "Density",
       fill = "State") + 
  theme_bw()

Heatplots with geom_tile()

You can create heatplots using the geom_tile() function. Try creating the following heatplot of statistics of NBA players using the following template:

# Read in nba data
nba_long <- read.csv("https://raw.githubusercontent.com/therbootcamp/therbootcamp.github.io/master/_slides/data/nba_long.csv")

ggplot(XX, 
       mapping = aes(x = XX, y = XX, fill = XX)) + 
  geom_tile(colour = "XX") + 
  scale_fill_gradientn(colors = c("XX", "XX", "XX"))+ 
  labs(x = "XX", 
       y = "XX", 
       fill = "XX", 
       title = "NBA XX performance",
       subtitle = "XX",
       caption = "XX") +
  coord_flip() + 
  theme_minimal()

# Read in nba data
nba_long <- read.csv("https://raw.githubusercontent.com/therbootcamp/therbootcamp.github.io/master/_slides/data/nba_long.csv")

ggplot(nba_long, 
       mapping = aes(x = Name, y = measure, fill = value)) + 
  geom_tile(colour = "white") + 
  scale_fill_gradientn(colors = c("red", "white", "blue"))+ 
  labs(x = "Player", 
       y = "Statistic", 
       fill = "Performance", 
       title = "NBA player performance",
       subtitle = "Each tile represents how well the player performed on that statistic relative to other players.",
       caption = "Source: https://learnr.wordpress.com/2010/01/26/ggplot2-quick-heatmap-plotting/") +
  coord_flip() + 
  theme_minimal()

Joyplots with ggjoy()

The ggjoy package contains a new geom called geom_joy() that creates a joyplot. Install the ggjoy package, load it, and then look at the introductory vignette by running vignette(topic = "introduction", package = "ggjoy"),

# Install the ggjoy package from CRAN
install.packages("ggjoy")

# Load the package
library("ggjoy")

# Open the introduction vignette
vignette(topic = "introduction", package = "ggjoy")

# Open the gallery vignette
vignette(topic = "gallery", package = "ggjoy")

Now, create the following joyplot of college education rates from the midwest data using the following template:

library(ggjoy)

ggplot(data = XX,
       mapping = aes(XX, y = XX, fill = XX)) + 
  geom_joy(col = "XX") + 
  scale_fill_cyclical(values = c("XX")) +
  labs(title = "XX",
       subtitle = "XX",
       caption = "XX",
       x = "XX",
       y = "XX") +
  scale_fill_brewer(palette = "XX") +
  theme_minimal()

library(ggjoy)

ggplot(data = midwest,
       mapping = aes(percollege, y = state, fill = state)) + 
  geom_joy(col = "white") + 
  scale_fill_cyclical(values = c("lightblue")) +
  labs(title = "College education rates",
       subtitle = "Wisconson, Ohio, Minnesota, Indiana, Illinois",
       caption = "Created with the ggjoy package",
       x = "Percent of residents with a college education",
       y = "State") +
  scale_fill_brewer(palette = "Set1") +
  theme_minimal()

Correlation plot with ggcorplot

The ggcorplot package lets you make really nice plots of correlation matrices. Install the package from github with the following code:

# Install the ggcorrplot package

#install.packages("devtools")   # If you don't have the devtools pacakge, you'll have to install it first
devtools::install_github("kassambara/ggcorrplot")

Using the following template, try to make your own correlation plot from the midwest data that looks like this:

• First create a correlation matrix using all of the numeric columns in the midwest data seen in the plot.
• Then, create the plot with ggcorrplot

library(ggcorrplot)

# Select only numeric rows
midwest_num <- midwest %>% select_if(is.numeric)

# Create a correlation matrix from several numeric columns
corr_mtx <- cor(midwest_num)

# Create the correlation plot!
ggcorrplot(cor = XX,
           type = "XX",
           lab = TRUE, 
           lab_size = 3, 
           method = "X", 
           colors = c("X", "X", "X"), 
           title = "XX",
           subtitle = "XX",
           caption = "XX",
           ggtheme = XX)

library(ggcorrplot)

# Select only numeric rows
midwest_num <- midwest %>% select_if(is.numeric)

# Create a correlation matrix from several numeric columns
corr_mtx <- cor(midwest_num)

# Plot
ggcorrplot(cor = corr_mtx,
           type = "lower",
           lab = TRUE, 
           lab_size = 3, 
           method = "circle", 
           colors = c("red", "white", "springgreen3"), 
           title = "Correlations of Measures from Midwest Counties",
           ggtheme = theme_bw)

Challenges

Make the following plot of savings data (psavert) from the economics dataset (hint: use the geom_line() function and map date to the x aesthetic and psavert to the y aesthetic.)

ggplot(data = XX, aes(x = XX, y = XX)) + 
       geom_line() + 
       labs(title = "XX", 
            subtitle = "XX", 
            caption = "XX", 
            y = "XX") + 
       theme_bw() + 
       geom_smooth()

# plot
ggplot(data = economics, 
       mapping = aes(x = date, y = psavert)) + 
  geom_line() + 
  geom_smooth() +
  labs(title = "Personal Savings Rates Changes over Time", 
       subtitle = "Ratio of personal saving to disposable income", 
       caption = "Source:  http://research.stlouisfed.org/fred2", 
       y = "Savings Rate %") + 
  theme_bw()

Make this plot from the diamonds data showing the relationship between diamond cut (cut), color (color), and carats (carat)

ggplot(data = diamonds, 
       aes(carat, fill = cut)) +
  facet_wrap(~ color, labeller = "label_both") + 
       geom_density(alpha = 0.5) + 
       labs(title = "Carats", 
            subtitle = "For diamonds of different cuts",
            caption = "Source: diamonds dataset in ggplot2",
            x = "Carats",
            y = "Density",
            fill = "Cut") + 
  theme_bw()

Make the following plot from the ACTG175 dataset. To do this, you'll need to use both geom_boxplot() and geom_point(). To jitter the points, use the position argument to geom_point(), as well as the position_jitter() function to control how much to jitter the points.

ggplot(data = ACTG175,
       aes(x = factor(arms), y = days, fill = factor(arms))) +
   facet_wrap(~ drugs, nrow = 1, labeller = label_both) +
  geom_boxplot(outlier.size = 0, alpha = .2) + 
  labs(title = "Number of days until a major negative event",
       subtitle = "For different treatment arms and separated by drug users and non drug users",
       caption = "Source: ACTG175 dataset from the speff2trial package",
       x = "Treatment Arm",
       y = "Number of days until a major negative event",
       fill = "Arm") +
  theme_bw() +  
  scale_fill_brewer(palette = "Dark2") + 
  geom_point(alpha = .2, 
             position = position_jitter(w = 0.1, h = 0))

The Davis dataframe in the car package contains measurements of the reported and measured (i.e.; real!) heights and weights of several men and women. Create the following plot showing the relationship between people's reported, and actual heights.

• Instead of using geom_point(), use geom_count() to create points whose size reflects the number of observations at that location.
• Include an "identity line" with the geom_abline() function.

library(car)

ggplot(data = Davis,
       mapping = aes(x = repht, y = height, color = sex)) + 
       geom_count(alpha = .8) + 
       scale_size_area(max_size = 10) + 
       geom_abline(slope = 1, intercept = 0) + 
       theme_bw() + 
  scale_y_continuous(limits = c(125, 200)) + 
  labs(x = "Reported height (cm)",
       y = "Measured height (cm)",
       title = "Relationship between reported and measured height",
       subtitle = "Separated by men and women",
       caption = "Source: Davis (1990). Body image and weight preoccupation")

You choose the plot!

ACTG175 (From the speff2trial package)

Create a plot showing the relationship between CD4 T cell count at 96 weeks (cd496), treatment arm (arms) and gender (gender).

Create a plot showing the relationship between treatment arm (arms), number of days until a major negative event (days) and history of intravenous drug use (drugs).

heartdisease (From the FFTrees package)

Create a plot showing the relationship between age, chol and diagnosis

Create a plot showing the relationship between cp, slope, and diagnosis

References

• Many of the plots in this practical were taken from Selva Prabhakaran's website http://r-statistics.co/Top50-Ggplot2-Visualizations-MasterList-R-Code.html

• For making maps with ggplot, check out Eric Anderson's tutorial at http://eriqande.github.io/rep-res-web/lectures/making-maps-with-R.html

therbootcamp/BaselRBootcamp2017 documentation built on May 3, 2019, 10:45 p.m.