Let’s visualize our data

Creating the base of the ggplot

To create our ggplot with nothing on it, we should specify two things:

1. The data set (fungicide)
2. The mapping of the x and y coordinates (from the data set, using aes)

Note, we can specify the column names without using quotation marks.

> yield.plot <- ggplot(data = fungicide, mapping = aes(x = Treatment, y = Yield_bu_per_acre))

If everything worked, you should see nothing. This is because ggplot2 returns an R object. This object contains the instructions for creating the visualization. When you print this object, the plot is created:

> yield.plot

Now you should see a plot with nothing on it where the x and y axes are labeled “Treatment” and “Yield_bu_per_acre”, respectively.

To break down what the above function did, it first took in the data set fungicide and then mapped the x and y aesthetics to the Treatment and Yield_bu_per_acre columns. Effectively, this told ggplot how big our canvas needs to be in order to display our data, but currently, it doesn’t know HOW we want to display our data; we need to give it a specific geometry.

Adding a geometry layer

All functions that add geometries to data start with geom_, so if we wanted the data to be displayed as a line showing the increase of yield over time, we would use geom_line(). If we wanted to show the data displayed as points, we can use geom_point().

To add a geometry or anything to a ggplot object, we can just use the + symbol. Here, we will add boxplots.

Note: From here on out, I will be wrapping all commands with parentheses. This allows the result of the assignment to be displayed automatically.

> #
> (yield.plot <- yield.plot + geom_boxplot())

If we want to change the color of the boxplots from white (default) to orange,
we can do this by adding geom_boxplot(fill = "orange").

> (yield.plot <- yield.plot + geom_boxplot(fill = "orange"))

Instead of all the boxplots having the same color, it will be interesting if we could color them according to the Treatment.

> yield.plot + geom_boxplot(fill = Treatment)

Error in layer(data = data, mapping = mapping, stat = stat, geom = GeomBoxplot, : object 'Treatment' not found

Oops! There was an error. It cannot recognize that we are talking about the Treatment column from our data set. This is because we have to use the function aes() whenever we are referring to our data set.

> (yield.plot <- yield.plot + geom_boxplot(aes(fill = Treatment))) # This works!

To give a title to our plot, we can use ggtitle().

> (yield.plot <- yield.plot + ggtitle("Effect of Fungicides on Yield"))

We now have a fully functional and informative plot using only few lines of code! Producing a visualization of your data can be an extremely useful tool for analysis because it can allow you to see if there are any strange patterns or spurious correlations in your variables.

We can click on ‘Zoom’ to view a bigger version of this plot.

Of course, this plot is not quite publication ready. We need to add some customization. Let’s manipulate the aesthetics of the plot in how the data and labels are displayed.

Axis labels

This is easily done with xlab() and ylab():

> (yield.plot <- yield.plot + xlab("Treatment Applied"))

> (yield.plot <- yield.plot + ylab("Yield (bu/acre)"))

The labels are now okay, but it’s still not publication-ready. The font is too small, the background should have no gridlines and the axis text needs to be darker. We also want to include the results of our ANOVA and Tukey’s HSD.

> #

Annotating with Tukey’s HSD letters

Remember that ‘Fungicide_B’ is signfificantly different from both ‘Fungicide_A’ and ‘Control’. ‘Fungicide_A’ and ‘Control’ are not signficantly different from each other. If we want to include stars or letters indicating significant differences we can use ‘annotate()’.
With Letters

> (yield.plot + annotate("text", x="Control", y= 174.5, label="A", size=5) +
+              annotate("text", x="Fungicide_A", y= 174.5, label="A", size=5)+
+              annotate("text", x="Fungicide_B", y= 176.25, label="B", size=5) )

With stars

> (yield.plot <- yield.plot + annotate("text", x="Fungicide_B", y= 176, label="*", size=10))

If we want to include a caption explaining what our annotations mean we can use ‘labs()’. We can also use this function to add subtitles, titles, or x and y labels.

> (yield.plot <- yield.plot + labs(caption="* indicates significant difference by Tukey's HSD pvalue <0.01"))

Adjusting Look and Feel (theme)

The first thing we can do is change the default theme from theme_grey() to theme_bw(). We will simultaneously set the base size of the font to be 14pt.

> (yield.plot <- yield.plot + theme_bw(base_size = 14))

There are many different default themes available for ggplot2 objects that change many aspects of the look and feel. The ggthemes contains many popular themes such as fivethirtyeight and economist. Of course, we can make it prettier before including it in our final manuscript.

To adjust granular aspects of the theme, we can use the theme() function, which contains a whopping 71 different options all related to the layout of the non-data aspects of the plot.

> stop("
+      Look at ?theme and figure out one of the following:
+      1. change the aspect ratio of the panels
+      2. remove the background grid in the panels
+      ")

Error in eval(expr, envir, enclos): 
     Look at ?theme and figure out one of the following:
     1. change the aspect ratio of the panels
     2. remove the background grid in the panels
     

When we inspect the help page of the theme() function, we can find out how to adjust several parameters to make out plot look acceptable:

> (yield.plot <- yield.plot + theme(aspect.ratio = 1)) # This looks the same

> (yield.plot <- yield.plot + theme(aspect.ratio = 2)) # This is too skinny

> (yield.plot <- yield.plot + theme(aspect.ratio = 1.25)) # I think this is perfect!

> (yield.plot <- yield.plot + theme(panel.grid = element_blank()))

Since the information in the legend is repetitive, we can remove it. If you ‘Google’ how to remove the legend in ggplot2, you will find that you can use guides(fill=FALSE).

> (yield.plot <- yield.plot + guides(fill=FALSE))

Putting it all together

Because we can add information to a plot with the + symbol, we can add all of the elements in one go. Let’s combine what we have above.

> yield.plot <- ggplot(fungicide, aes(x = Treatment, y = Yield_bu_per_acre)) +
+   geom_boxplot(aes(fill = Treatment)) +
+   ggtitle("Effect of Fungicides on Yield") +
+   xlab("Treatment Applied") +
+   ylab("Yield (bu/acre)") +
+   annotate("text", x="Fungicide_B", y= 176, label="*", size=10) +
+   labs(caption="* indicates significant difference by Tukey's HSD <0.01") +
+   theme_bw(base_size = 14) +
+   theme(aspect.ratio = 1.25) +
+   theme(panel.grid = element_blank()) +
+   guides(fill=FALSE)
> yield.plot

We can now create a similar plot for Severity Data.

> severity.plot <- ggplot(fungicide, aes(x = Treatment, y = Severity)) +
+   geom_boxplot() +
+   ggtitle("Effect of Fungicides on Disease Severity") +
+   theme_bw(base_size = 14) +
+   theme(aspect.ratio = 1.5) +
+   theme(panel.grid = element_blank()) +
+   xlab("Treatment Applied") +
+   ylab("Disease Severity") +
+   annotate("text", x="Control", y= 5.75, label="a", size=5) +
+   annotate("text", x="Fungicide_A", y= 5.35, label="b", size=5)+
+   annotate("text", x="Fungicide_B", y= 4.35, label="c", size=5) +
+   labs(caption="letters indicates significant differences by Tukey's HSD <0.01") 
> severity.plot

The text of the title is not in the center. To format text elements of the plot, we can use the function element_text() inside theme(). Since we need to edit the text of the plot title, we need to specify plot.title = element_text(). Since we need to adjust the text horizontally in the center, we modify it to theme(plot.title = element_text(hjust = 0.5)).

> (severity.plot <- severity.plot + theme(plot.title = element_text(hjust = 0.5), plot.caption=element_text(hjust=.5)))