Visualization often plays a major role in the research process, from quality assurance, to data exploration, to the presentation of results.

The data visualization package ggplot2 makes it very easy and straightforward to create a lot of different types of plots, from simple to complex. In this module, we’ll introduce ggplot syntax and briefly survey some of the package’s plotting capabilities.

A good ggplot cheat sheet can be found here: ggplot cheat sheet

Load script for module #1.3: Plotting

  1. Click here to download the script, and save the script to your working directory (R Project folder).

  2. Load the script into your RStudio Project. To do this, open your RStudio Project and click on the folder icon in the toolbar at the top and load your script.

Let’s get started with plotting in R!

ggplot basics

A typical workflow begins with initiating plotting with the ggplot() function and specifying the data frame you want to use to create your visualizations. We then often define the “x” argument (defining coordinates on an x axis) and (if applicable) a “y” argument (defining coordinates on a y axis). In ggplot these are known as aesthetic mappings! That is because it is a way of conveying information in the dataset graphically! Other ways of conveying information graphically include point size, color, and symbol- all of which are also referred to as aesthetic mappings (syntax: mapping=aes()).

You can then add geometric objects (like points) to your plot, often using functions beginning with geom_, to represent your data in the form of a boxplot (geom_boxplot()), scatterplot (geom_point()), or a variety of other types of plots. The aes() function can be called within these “geoms” to specify which variables to display and how these data should be displayed (which data should be used to represent the bar height, or the x coordinate, or the point size, etc.). aes() can also be used within the initial call to ggplot().

Here is an example of a simple scatterplot in ggplot:

# scatterplots ----------------------------

ggplot(trees, aes(x=Girth,y=Volume)) +
  geom_point()

We’ll start with the ‘trees’ dataset, which is built into R. It describes the girth, height, and volume of 31 felled black cherry trees.

# ASIDE: explore the built-in 'trees' dataset

# ?trees      # description of built in dataset  (uncomment to run)

dim(trees)   # Show the dimension of the trees dataframe                                              
## [1] 31  3
str(trees)   # Show the structure of the trees dataframe
## 'data.frame':    31 obs. of  3 variables:
##  $ Girth : num  8.3 8.6 8.8 10.5 10.7 10.8 11 11 11.1 11.2 ...
##  $ Height: num  70 65 63 72 81 83 66 75 80 75 ...
##  $ Volume: num  10.3 10.3 10.2 16.4 18.8 19.7 15.6 18.2 22.6 19.9 ...
head(trees)   # Show the first few observations of the trees dataframe
##   Girth Height Volume
## 1   8.3     70   10.3
## 2   8.6     65   10.3
## 3   8.8     63   10.2
## 4  10.5     72   16.4
## 5  10.7     81   18.8
## 6  10.8     83   19.7
summary(trees)  # Summary stats for each column
##      Girth           Height       Volume     
##  Min.   : 8.30   Min.   :63   Min.   :10.20  
##  1st Qu.:11.05   1st Qu.:72   1st Qu.:19.40  
##  Median :12.90   Median :76   Median :24.20  
##  Mean   :13.25   Mean   :76   Mean   :30.17  
##  3rd Qu.:15.25   3rd Qu.:80   3rd Qu.:37.30  
##  Max.   :20.60   Max.   :87   Max.   :77.00

And some more fancy and informative scatterplots

    # try representing tree height using the color aesthetic
ggplot(trees, aes(x=Girth,y=Volume)) +
  geom_point(aes(col=Height))

    # try representing tree height using the size aesthetic
ggplot(trees, aes(x=Girth,y=Volume)) +
  geom_point(aes(size=Height))

   # try adding a regression line
ggplot(trees, aes(x=Girth,y=Volume)) +
  geom_point() +
  geom_smooth(method="lm")

Change Plot Type

Because we’re exploring different ways of plotting, it is useful to include multiple plots in the same image.

We can do this using the plot_grid() function in the cowplot package (commonly used extensions to ggplot). Here we explore different ways of graphically representing the relationship between tree girth and volume.

# Explore different "geoms" or plot types 

plot1 <- ggplot(trees,aes(Girth,Volume)) +    # plot the relationship as a line
  geom_line()
plot2 <- ggplot(trees,aes(Girth,Volume)) +    # plot a smoothed "spline" fit of the relationship
  geom_smooth()
plot3 <- ggplot(trees,aes(Girth,Volume)) +    # plot scatterplot
  geom_point() 
plot4 <- ggplot(trees,aes(Girth,Volume)) +    # plot scatterplot with smoothed regression line
  geom_point() + 
  geom_smooth()
plot_grid(plot1,plot2,plot3,plot4,labels="auto")

Change symbols, colors and point sizes

Point shape is specified using the shape= option.

Color is specified using the color= option.

Size is specified using the size= option.

If you want convey information in your data via the color, symbol, or point size, you include it as an aesthetic mapping- and link it to one of the columns in your data.

If you just want to color all the points the same way, or use the same symbol for everything, specify this outside of the aesthetic mapping.

To illustrate this, let’s use the built in iris dataset, which gives the sepal length/width and petal length/width (in cm), for 50 flowers from each of 3 iris species.

# Explore different aesthetic mappings 

plot1 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # shape represents species
  geom_point(aes(shape=Species))
plot2 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +       # shape is same for everything!
  geom_point(shape=2)
plot3 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) 
plot4 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color is same across the board
  geom_point(col="purple")  
plot5 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # size represents species
  geom_point(aes(size=Species)) 
plot6 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # size is same across the board
  geom_point(size=2) 
plot_grid(plot1,plot2,plot3,plot4,plot5,plot6,ncol=2)

Here is a graphic illustrating the main plot symbols you can reference using shape=:

Colors are a bit more intuitive, but here’s a link to a chart of R colors)

Themes

You can see that the default plot includes a gray background with white gridlines. This makes all of the elements on this plot easy to see, but as you start adjusting colors and identifying your personal preferences, you’ll probably want to customize this – ggplot has a ton of options for doing so. Here’s a few examples of ggplot themes:

# explore themes 

plot1 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) +
  theme_bw()
plot2 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) +
  theme_classic()
plot3 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) +
  theme_minimal()
plot4 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) +
  theme_minimal_grid(font_size = 11)

plot_grid(plot1,plot2,plot3,plot4,labels = "AUTO")

# note: many other themes are available in ggplot, cowplot and other related packages

Changing title and axis labels

We can also add titles, axis labels, and other options to make the plots look prettier.

# add additional plot elements: title, axis limis, axis labels 

plot1 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) + 
  labs(x="Sepal Length (cm)")

plot2 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) + 
  labs(x="Sepal Length (cm)",y="Petal Length (cm)",color="Iris sp.")

plot3 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) + 
  labs(x="Sepal Length (cm)",y="Petal Length (cm)",color="Iris sp.",
       title="Fisher's Iris Data",subtitle = "practice with ggplot")

plot4 <- ggplot(iris,aes(Sepal.Length,Petal.Length)) +    # color represents species
  geom_point(aes(color=Species)) + 
  coord_cartesian(xlim=c(0,10),ylim=c(0,10)) +
  labs(x="Sepal Length (cm)",y="Petal Length (cm)",color="Iris sp.",
       title="Fisher's Iris Data",subtitle = "practice with ggplot")

plot_grid(plot1,plot2,plot3,plot4,labels = "AUTO")

Bar Plots and box-whisker plots

# bar plots, box-whisker plots, violin plots  -----------------------

plot1 <- ggplot(iris,aes(x=Species,y=Sepal.Length)) +    # more informative box-whisker plot
  geom_boxplot() 
  
plot2 <- ggplot(iris,aes(x=Species,y=Sepal.Length)) +    # more informative box-whisker plot +
  geom_violin() 

# bar plot
bar.heights <- iris %>% 
  group_by(Species) %>% 
  summarize(meanSL = mean(Sepal.Length))

plot3 <- ggplot(bar.heights, aes(Species,meanSL)) +
  geom_col()


plot4 <- ggplot(bar.heights, aes(Species,meanSL)) +
  geom_col(aes(fill=Species)) +
  theme_classic() +
  scale_fill_manual(values=c("gray","red","brown"))
  
plot_grid(plot1,plot2,plot3,plot4,labels = "AUTO")

But what if we want to have some error bars?

# Bar plot with error bars

bar.heights <- iris %>% 
  group_by(Species) %>% 
  summarize(meanSL = mean(Sepal.Length),
            n = n(),
            sdSL = sd(Sepal.Length),
            se = sdSL/sqrt(n))
  
ggplot(bar.heights,aes(x=Species,y=meanSL)) + 
  geom_col(fill=gray(0.7),color="black") +
  geom_errorbar(aes(ymin=meanSL-2*sdSL,ymax=meanSL+2*sdSL),width=.2) +
  labs(y="Sepal Length")

And for a slightly more complex example, let’s consider the built in ToothGrowth data set, which looks at tooth growth in guinea pigs under three different vitamin C doses and two different delivery methods (orange juice or ascorbic acid).

# ?ToothGrowth
head(ToothGrowth)
##    len supp dose
## 1  4.2   VC  0.5
## 2 11.5   VC  0.5
## 3  7.3   VC  0.5
## 4  5.8   VC  0.5
## 5  6.4   VC  0.5
## 6 10.0   VC  0.5
# toothgrowth plot 


ToothGrowth$dose <- as.factor(ToothGrowth$dose)

sumTC <- ToothGrowth %>% 
  group_by(supp,dose) %>% 
  summarize(mean = mean(len),
            sd = sd(len))


p<- ggplot(sumTC, aes(x=dose, y=mean, fill=supp)) + 
  geom_col(color="black", 
           position=position_dodge()) +
  geom_errorbar(aes(ymin=mean-sd, ymax=mean+sd), width=.2,
                 position=position_dodge(0.9)) +
  labs(title="Tooth growth", x="Dose (mg)", y = "Length") +
   theme_classic() +
   scale_fill_manual(values=c('#999999','#E69F00'))

print(p)

Beyond ggplot

ggplot contains an incredibly rich and powerful set of tools for visualizing data. But ggplot and base R are not your only options! A variety of packages exist for visualization, including:

ggplot extensions

Several packages have been created that build off of ggplot’s syntax with additional functions. You can find a list of them here.

Interactive plots: leaflet and plotly

Increasingly, scientific journals are providing platforms for interactive graphics on their websites to accompany published articles. Interactive plots are also popular for personal, lab, and organizational websites, and they can provide another option for your own data exploration. Two of the most popular in R are plotly, which offers a huge variety of 2D and 3D plots, and leaflet, which is specifically for mapping.

Here’s a quick and simple example of leaflet in action using the “possumsites” dataframe that accompanies the “possum” dataset in the DAAG package. The dataset contains body measurements of several possums in Australia. Where did they catch these possums?

# use leaflet for interactive mapping! ------------

leaflet(possumsites) %>%
  addTiles() %>% #Adds map tiles from OpenStreetMap
  addMarkers(lng=c(possumsites$Longitude), lat=c(possumsites$Latitude), 
             popup=c(as.character(possumsites$altitude))) #Adds markers for the sites

Challenge exercises

  1. Using the built in ‘mtcars’ dataset, make a scatterplot of ‘mpg’ as a function of ‘disp’. Color the points according to the ‘cyl’ variable, and make the size of points vary depending on the ‘hp’ variable. Try to code the ‘cyl’ variable as a categorical variable (factor). The plot should look something like this:

  1. Again using the ‘mtcars’ data, make a scatterplot of ‘mpg’ as a function of ‘disp’. Color the points according to the ‘cyl’ variable, and make a separate trendline for each unique value of the ‘cyl’ variable (with trendlines also indicating the ‘cyl’ variable with the appropriate color).

# CHALLENGE EXERCISES   -------------------------------------

# 1. Using the built in 'mtcars' dataset, make a scatterplot of 'mpg' as a function of 'disp'. Color the points
# according to the 'cyl' variable, and make the size of points vary depending on the 'hp' variable. Try to code
# the 'cyl' variable as a categorical variable (factor). The plot should look something like this:
#
# 2. Again using the 'mtcars' data, make a scatterplot of 'mpg' as a function of 'disp'. Color the points
# according to the 'cyl' variable, and make a separate trendline for each unique value of the 'cyl' variable (with
# trendlines also indicating the 'cyl' variable with the appropriate color).

–go to next submodule–

More advanced GGplot demo (optional!)

For additional practice, let’s use the Soils dataset from the ‘carData’ package, which contains soil attributes from a gilgai landscape (mounds and depressions caused by shrinking and swelling of clays during dry and wet seasons) in Australia.

First we will need to load a few more packages. Make sure to install any packages in this list you don’t already have!

# More complex demo -----------------------------

library(ggthemes)
library(carData)
library(DAAG)
library(RColorBrewer)

Now load the dataset.

# Load the example data 

soil <- carData::Soils    # load example data

#See what variables it contains...
head(soil)    # plot out the first few lines...
##   Group Contour Depth Gp Block   pH     N Dens   P    Ca   Mg    K   Na Conduc
## 1     1     Top  0-10 T0     1 5.40 0.188 0.92 215 16.35 7.65 0.72 1.14   1.09
## 2     1     Top  0-10 T0     2 5.65 0.165 1.04 208 12.25 5.15 0.71 0.94   1.35
## 3     1     Top  0-10 T0     3 5.14 0.260 0.95 300 13.02 5.68 0.68 0.60   1.41
## 4     1     Top  0-10 T0     4 5.14 0.169 1.10 248 11.92 7.88 1.09 1.01   1.64
## 5     2     Top 10-30 T1     1 5.14 0.164 1.12 174 14.17 8.12 0.70 2.17   1.85
## 6     2     Top 10-30 T1     2 5.10 0.094 1.22 129  8.55 6.92 0.81 2.67   3.18

Staring simply, let’s say we’re interested in how pH, a continuous variable, varies with contour position, a categorical factor:

# basic boxplot and violin plot

plot1 <- ggplot(soil) +
  geom_boxplot(aes(x=Contour, y=pH))

plot2 <- ggplot(soil) + 
  geom_violin(aes(x=Contour, y=pH))

plot_grid(plot1,plot2,labels = "AUTO")

Next let’s try a simple scatterplot of two continuous variables, Calcium content and pH:

# basic scatterplot

ggplot(soil) +
  geom_point(aes(x=pH, y=Ca))

Now let’s say you wanted to examine if/how the relationship between pH and Ca varies with depth. You could go back to your scatterplot and use color to identify points from the different sampling depths.

# Color the points by depth

ggplot(soil) +
  geom_point(aes(x=pH, y=Ca, color=Depth))

Try playing around with applying a few graphical attributes to all the points (specified outside the aes() argument). Try specifying shape=21, color="black" , size=3 and stroke=1:

# make additional alterations (outside the "aes" function)

ggplot(soil) +
  geom_point(aes(x=pH, y=Ca, fill=Depth), shape=21, color="black", size=4, stroke=1.5)

Now for something new: let’s plot relationships between pH and several soil nutrients on the same graph. Notice that the y axis label defaults to the first set of points (Ca), and we’d need to modify it. Also there is no legend! You’d have to change things like axis labels and legend titles ‘manually’ to complete this figure.

# Plot several relationships on same plot

ggplot(soil, aes(x=pH)) +
  geom_point(aes(y=Ca), shape=21, fill="red", color="black", size=4, stroke=1.5) +
  geom_point(aes(y=Mg), shape=21, fill="blue", color="black", size=4, stroke=1.5) +
  geom_point(aes(y=Na), shape=21, fill="gray30", color="black", size=4, stroke=1.5)

There’s a better way to do it. But first we need to ‘reshape’ our soils data with the tidyverse function `pivot_longer().

ASIDE: tidy data and reshaping functions

All of the tidyverse set of packages are designed to work with Tidy formatted data.

This means:

  • Each variable must have its own column.
  • Each observation must have its own row.
  • Each value must have its own cell.

This is what it looks like:

The tidyr package has several functions to help you get your data into this format.

Use “pivot_longer()” to get all of the values and variables from multiple columns into a single column.

Use “pivot_wider()” to distribute two variables in a single column into separate columns, with their data values(‘value’)

For a nice intro to ‘pivot_wider’ and ‘pivot_longer’, see this link

Back to the example…

Let’s reshape the soil dataset into the long format- which in this case treats each separate nutrient measurement as a different observation (row).

# Use 'tidyverse' to reshape the data 

soil.nut <- pivot_longer(soil, cols=c("Ca","Mg","Na"), names_to="nutrient",values_to = "value" )
soil.nut
## # A tibble: 144 × 13
##    Group Contour Depth Gp    Block    pH     N  Dens     P     K Conduc nutrient
##    <fct> <fct>   <fct> <fct> <fct> <dbl> <dbl> <dbl> <int> <dbl>  <dbl> <chr>   
##  1 1     Top     0-10  T0    1      5.4  0.188  0.92   215  0.72   1.09 Ca      
##  2 1     Top     0-10  T0    1      5.4  0.188  0.92   215  0.72   1.09 Mg      
##  3 1     Top     0-10  T0    1      5.4  0.188  0.92   215  0.72   1.09 Na      
##  4 1     Top     0-10  T0    2      5.65 0.165  1.04   208  0.71   1.35 Ca      
##  5 1     Top     0-10  T0    2      5.65 0.165  1.04   208  0.71   1.35 Mg      
##  6 1     Top     0-10  T0    2      5.65 0.165  1.04   208  0.71   1.35 Na      
##  7 1     Top     0-10  T0    3      5.14 0.26   0.95   300  0.68   1.41 Ca      
##  8 1     Top     0-10  T0    3      5.14 0.26   0.95   300  0.68   1.41 Mg      
##  9 1     Top     0-10  T0    3      5.14 0.26   0.95   300  0.68   1.41 Na      
## 10 1     Top     0-10  T0    4      5.14 0.169  1.1    248  1.09   1.64 Ca      
## # ℹ 134 more rows
## # ℹ 1 more variable: value <dbl>

This lets us plot our different nutrients as factor levels without typing each one with its specifications on its own line, and automatically generates a legend that identifies the series (and also gives the default ggplot colors).

ggplot(soil.nut) +
  geom_point(aes(x=pH, y=value, fill=nutrient), shape=21, color="black", size=4, stroke=1.5)

What if instead of Na, we had K as our third nutrient of interest?

# or if we wanted to plot different nutrients...

soil.nut2 <- pivot_longer(soil, cols=c("Ca","Mg","K"), names_to="nutrient",values_to = "value" )

ggplot(soil.nut2) +
  geom_point(aes(x=pH, y=value, fill=nutrient), shape=21, color="black", size=4, stroke=1.5)

The problem here is that K occurs at natural concentrations that differ from the other nutrients by an order of magnitude, so it’s difficult to examine them all on the same scale. One approach we can take is to plot them separately. While we’re at it, let’s also make sure that we specify the response units on the y-axis.

Facets and scales, and themes

The next plot introduces:

  • facets, which allow you to display data on separate panels using some grouping variable

  • scales, which allow you to adjust how to represent the data with axes and colors

We also play around with some more ggplot themes.

# plot with facets, scales, and themes!

ggplot(soil.nut2) +
  geom_point(aes(x=pH, y=value, fill=nutrient), 
             shape=21, color="black", size=4, stroke=1.5) +
  facet_wrap(~nutrient, scales="free_y") +
  labs(y="mg / 100 g soil") +
  theme_bw() +
  theme(legend.position="none",
        axis.text = element_text(size=20),
        axis.title = element_text(size=25),
        strip.text = element_text(size=25, face="bold"))

In this plot, we’ve kept the different colors for each nutrient, but suppressed the legend that is auto-generated by ggplot, because this information is now redundant with the headers of the three facets. In the facet_wrap function, we used scales to specify that the y-axis should be bounded to fit the data within each facet. There are some cases where you would probably want the axes to be the same – e.g. if you were comparing raw values across groups, rather than the distributions or trends of the data.

We also changed the theme. ggplot has 8 built-in themes to choose from. There are also lots of other themes built into extension packages such as ggthemes.

You can also fine-tune pretty much any of the details (gridlines or no? plot borders? line weight of the borders and gridlines? axis label angles?) to your heart’s content using the arguments in the theme() function.

ASIDE: Specifying colors

Sometimes sticking to the default colors in ggplot isn’t the best choice. You might have factors representing ordered ranks, like in an experiment with different levels of light exposure, and want to represent these levels on a monochromatic scale. Or you might want to make a map displaying regions of positive or negative change in forest cover, using a diverging color scale. Or you don’t even like the default ggplot colors, and have your own preferred color schemes. It’s also important to remember that red-green colorblindness is fairly common, so if you are presenting data that must be distinguished by colors in a single plotting area, you should probably avoid this combination or combine it with changes in value (light to dark) in order for your plot to be accessible.

Going back to our earlier example showing Ca content by pH at different depths, let’s say we want a color scheme where deeper depths are represented by darker values of the same color. We can do this by using another scale function.

One method is to use scale_fill_brewer, and select an already composed color palette from RColorBrewer (a package you’ll need to install). You can check out all of the options available in RColorBrewer by entering display.brewer.all(), which shows the sequential palettes, then categorical palettes, than diverging palettes.

# Playing with colors in ggplot!

display.brewer.all()

I’m going to pick the YlOrBr (Yellow-Orange-Brown) palette, because those seem like good soil-y colors. Notice that because fill is mapped to the values of the data, inside this function is where I can change the title of the legend (and the labels for the different values, if I wanted to do that).

# Choose a new color palette from the RColorBrewer package

ggplot(soil) +
  geom_point(aes(x=pH, y=Ca, fill=Depth), shape=21, color="black", size=4, stroke=1.5) +
  theme_classic() +
  labs(y="Ca (mg/100g soil)") +
  scale_fill_brewer(palette="YlOrBr", name="Depth (cm)")

Another method is to use scale_fill_manual, with which you define your own palette. To pick out your colors, you can use the names of colors already recognized by R, or use hex codes for any color you want.

# Choose your own palette!

ggplot(soil) +
  geom_point(aes(x=pH, y=Ca, fill=Depth), shape=21, color="black", size=4, stroke=1.5) +
  theme_bw() +
  ylab("Ca (mg/100g soil)") +
  scale_fill_manual(values=c("#FFF0BF","#FFC300","#BF9200","#604900"), name="Depth (cm)")

What if depth were a continuous variable, rather than a set of four discrete categories? We could use scale_fill_gradient, supply the hues we want (one for monochromatic, two for diverging) and R would map the soil depth values to color values along the gradient.

Trendlines

Next, we might want to add a trendline to each set of points. Try it for the plot we made of different nutrient concentrations as a function of pH. Arguments in the geom_smooth() function allow us to change the confidence level, the smoothing method, and other details:

# add trendlines

ggplot(soil.nut2) +
  geom_point(aes(x=pH, y=value, fill=nutrient), 
             shape=21, color="black", size=4, stroke=1.5) +
  geom_smooth(aes(x=pH, y=value), method="lm", color="black") +
  facet_wrap(~nutrient, scales="free_y") +
  ylab("mg / 100 g soil") +
  theme_classic() +
  theme(legend.position="none",
        axis.text = element_text(size=14),
        axis.title = element_text(size=16),
        strip.text = element_text(size=16, face="bold"))

Density curves and error bars

Using geom_smooth to plot trendlines is a useful way to get more information from your scatterplots. To get an idea for how other kinds of statistical elements can be drawn onto plots, we’ll look at examples using histograms and boxplots.

First, a simple histogram of 3 soil nutrients.

# Adding density/smooth curves to plots

   ## first produce some histograms

ggplot(soil.nut) +
  geom_histogram(aes(x=value), color="black", fill="white", bins=15) +
  facet_wrap(~nutrient, scales="free") +
  xlab("mg / 100g soil") +
  theme_classic() +
  theme(axis.text = element_text(size=20),
        axis.title = element_text(size=25),
        strip.text = element_text(size=25, face="bold"))

Here is how to add a density curve to the histogram with another geom:

# Then add density curves

ggplot(soil.nut) +
  geom_histogram(aes(x=value, y=..density..), color="black", fill="white", bins=15) +
  geom_density(aes(x=value,color=nutrient), size=1.5) +
  facet_wrap(~nutrient, scales="free") +
  xlab("mg / 100g soil") +
  theme_classic() +
  theme(legend.position="none",
        axis.text = element_text(size=20),
        axis.title = element_text(size=25),
        strip.text = element_text(size=25, face="bold"))

And here’s how we can compare our distributions to a normal distribution, using a stat function:

# And now let's use a statistical function (dnorm) in ggplot to compare with a normal distribution:

ggplot(soil.nut) +
  geom_histogram(aes(x=value, y=..density..), color="black", fill="white", bins=15) +
  stat_function(fun = dnorm, color = "blue", size = 1.5,
                args=list(mean=mean(soil.nut$value), sd=sd(soil.nut$value))) +
  facet_wrap(~nutrient, scales="free") +
  xlab("mg / 100g soil") +
  theme_classic() +
  theme(legend.position="none",
        axis.text = element_text(size=20),
        axis.title = element_text(size=25),
        strip.text = element_text(size=25, face="bold"))

Let’s go back to our basic boxplot, the default plot for visualizing continuous data against categories.

First, you may have noticed that the whiskers do not have caps on them, as they do in base R. This is a matter of personal preference, but if capless whiskers look “off” to you, then you can work around it using stat_boxplot. stat_ functions are another group of functions for creating layers based on statistical properties of the data. Another example is stat_summary, which we use in the example below to add means to the boxplot.

# add error bars and other stat summaries (e.g., mean) to boxplot

ggplot(soil, aes(x=Contour, y=pH)) +
  stat_boxplot(geom="errorbar", width=0.2) +  
  stat_boxplot() +
  stat_summary(fun=mean,geom="point",size=5, color="black")

Interestingly, there is generally a way to add just about any graphical element using either “stat_” or “geom_” functions. It’s a good idea to play around with both!

Technique-specific plotting libraries

Several analytical packages come with their own plotting functions that produce some very nice visualizations. There are dozens out there, but a few of them are visreg for regression plots, corrplot (and its ggplot counterpart, ggcorrplot) for graphical presentation of correlation matrices, and rpart.plot as a companion to the decision tree package rpart.