ggplot2: Plotting with Variable Variables

.Rmd file is available at https://github.com/meekj/TRU-July-2020

Problem

You have a data frame with more than two columns that you want to plot.
You want to minimize repeated ggplot code blocks
You want to quickly pick pairs of variables to plot
You want to stack plots with different pairs of variables
You need to auto-generate a report where the number and type of plots is data driven

Solution

We will use aes_string
- Note that aes_string is "soft deprecated" as of ggplot2 3.0.x
- Tidy evaluation is the future, for another time...
Plots will be stacked using ggpubr::ggarrange
- ggarrange keeps alignment when y-axis label space varies

Discussion

I was going to use York River wind data
But palmerpenguins is small and easily accessible
- Not yet on CRAN
- devtools::install_github('allisonhorst/palmerpenguins')
Previously, I used get() to implement plotting variable variables but it seems to be a hack with issues when used for this purpose

Initialization

## To run from command line:
## Rscript  -e "rmarkdown::render('~/lab/R/var-var-plot.Rmd', output_dir = '~/r-reports/')"

library(dplyr)
library(ggplot2)
library(ggpubr)          # Stack plots with ggpubr::ggarrange
library(palmerpenguins)  # Sample data

pointSize <- 2

theme_jm1 <- theme_bw() + # A decent theme for HTML output
    theme(
        plot.title  = element_text(size = rel(1.5), family = 'Helvetica', face = 'bold'),
        plot.subtitle  = element_text(size = rel(1.3), family = 'Helvetica', face = 'bold'),
        axis.title  = element_text(size = rel(1.5), colour = 'black', face = 'bold'),
        axis.text.x = element_text(angle=0, size = rel(1.5), lineheight = 0.9, colour = 'black', vjust = 1, face = 'bold'),
        axis.text.y = element_text(size = rel(1.5), lineheight = 0.9, colour = 'black', hjust = 1, face = 'bold'),
        legend.title = element_text(size = rel(1.75)),
        legend.key  = element_rect(colour = 'white', fill = 'white'),
        legend.text = element_text(size = rel(1.3))
    )

FigureWidth  <- 14
FigureHeight <- 9

str(penguins)

## tibble [344 × 7] (S3: tbl_df/tbl/data.frame)
##  $ species          : Factor w/ 3 levels "Adelie","Chinstrap",..: 1 1 1 1 1 1 1 1 1 1 ...
##  $ island           : Factor w/ 3 levels "Biscoe","Dream",..: 3 3 3 3 3 3 3 3 3 3 ...
##  $ bill_length_mm   : num [1:344] 39.1 39.5 40.3 NA 36.7 39.3 38.9 39.2 34.1 42 ...
##  $ bill_depth_mm    : num [1:344] 18.7 17.4 18 NA 19.3 20.6 17.8 19.6 18.1 20.2 ...
##  $ flipper_length_mm: int [1:344] 181 186 195 NA 193 190 181 195 193 190 ...
##  $ body_mass_g      : int [1:344] 3750 3800 3250 NA 3450 3650 3625 4675 3475 4250 ...
##  $ sex              : Factor w/ 2 levels "female","male": 2 1 1 NA 1 2 1 2 NA NA ...

Code Example 1

## For interactive use select a set of variables, batch job uses the last set defined

vars_to_plot_x <- c('body_mass_g', 'body_mass_g', 'body_mass_g')
vars_to_plot_y <- c('bill_length_mm', 'bill_depth_mm', 'flipper_length_mm')

vars_to_plot_x <- c('flipper_length_mm', 'bill_length_mm', 'bill_depth_mm')
vars_to_plot_y <- c('bill_length_mm', 'bill_depth_mm', 'flipper_length_mm')

penguinPlots <- list()                     # Holds individual plots until they are stacked with ggarrange
for (varnum in 1:length(vars_to_plot_x)) { # Variable variable names
    Title <- paste0('Palmer Penguins', ' - ', vars_to_plot_x[[varnum]], ' - ', vars_to_plot_y[[varnum]])
    penguinPlots[[varnum]] <- ggplot(penguins) +
        geom_point(aes_string(x = vars_to_plot_x[[varnum]], y = vars_to_plot_y[[varnum]], colour = 'species'), size = pointSize) +
        xlab(vars_to_plot_x[[varnum]]) + ylab(vars_to_plot_y[[varnum]]) +
        ggtitle(Title)  + theme_jm1
}

Penguins, penguins, penguins

ggarrange(plotlist = penguinPlots, ncol = 1, nrow = length(penguinPlots), align = "v")

Make it nicer, and more report generator friendly

vars_to_plot_x <- c('body_mass_g', 'body_mass_g', 'body_mass_g')
vars_to_plot_y <- c('bill_length_mm', 'bill_depth_mm', 'flipper_length_mm')

axis_labels_x <- c('', '', 'Body Mass, g')
axis_labels_y <- c('Bill Length, mm', 'Bill Depth, mm', 'Flipper Length, mm')

genPlot <- function(data, title, vars_x, vars_y, ax_lbls_x, ax_lbls_y, var_color) {
    plots <- list()
    for (varnum in 1:length(vars_x)) {
        plots[[varnum]] <- ggplot(data) +
            geom_point(aes_string(x = vars_x[[varnum]], y = vars_y[[varnum]], colour = var_color), size = pointSize) +
            xlab(ax_lbls_x[[varnum]]) + ylab(ax_lbls_y[[varnum]]) +
            theme_jm1
        if (varnum == 1) plots[[varnum]] <- plots[[varnum]] + ggtitle(title) # Title only on top plot
    }
    return(plots)
}

Generate and display a plot set - colorized by species

## Note that all plots are vertically aligned regardless of y-axis tick digits

penguinPlots2 <- genPlot(penguins,'Palmer Penguins', vars_to_plot_x, vars_to_plot_y, axis_labels_x, axis_labels_y, 'species')
ggarrange(plotlist = penguinPlots2, ncol = 1, nrow = length(penguinPlots2), align = "v")

How about labeling the X axis only on the bottom plot?

genPlot2 <- function(data, title, vars_x, vars_y, ax_lbls_x, ax_lbls_y, var_color) {
    plots <- list()
    for (varnum in 1:length(vars_x)) {
        plots[[varnum]] <- ggplot(data) +
            geom_point(aes_string(x = vars_x[[varnum]], y = vars_y[[varnum]], colour = var_color), size = pointSize) +
            xlab(ax_lbls_x[[varnum]]) + ylab(ax_lbls_y[[varnum]]) +
            theme_jm1
        if (varnum == 1)              plots[[varnum]] <- plots[[varnum]] + ggtitle(title)                    # Title only on top plot
        if (varnum != length(vars_x)) plots[[varnum]] <- plots[[varnum]] + scale_x_continuous(labels = NULL) # X axis labels only on bottom
    }
    return(plots)
}

Generate and display a plot set - colorized by sex

penguinPlots2 <- genPlot2(penguins,'Palmer Penguins', vars_to_plot_x, vars_to_plot_y, axis_labels_x, axis_labels_y, 'sex')
ggarrange(plotlist = penguinPlots2, ncol = 1, nrow = length(penguinPlots2), align = "v")

Conclusion and Notes

These techniques provide an easier method of producing dynamic reports, where the number of plots is determined by the data
They should also aid interactive exploratory plotting
ggplot objects "may" be rendered only at print time
- Actually this is probably always true but the rendering point might not be obvious
- This could result in the expected labels but only the last defined data set
- We do not have this problem with the above examples

ggplot2: Plotting with Variable Variables

Jon Meek - Trenton R Users

18 July 2020

Problem

Solution

Discussion

Initialization

Code Example 1

Penguins, penguins, penguins

Make it nicer, and more report generator friendly

Generate and display a plot set - colorized by species

How about labeling the X axis only on the bottom plot?

Generate and display a plot set - colorized by sex

Conclusion and Notes