Plotting2.md

Plotting: week 2

author: date: autosize: true

Outline

• Advanced layout
• Symbols (avoiding "expression/bquote hell")
• Papers vs talks vs posters
• Automated reports
• Python plotting (may have to wait till lab/next week)

Advanced layout

• Multi-panel plots are in every paper.
• Where possible, I prefer to make each panel separately, so that LaTeX's subfigure environment allows cross-referencing.
• However, that strategy means MORE graphics files to upload to a journal, which sucks.
• So--panels, how to do them?

Cowplot

library(cowplot)
# library(ggplot2)

==============================

plot.mpg <- ggplot(mpg, aes(x = cty, y = hwy, colour = factor(cyl))) + 
  geom_point(size=2.5)
plot.mpg

==================

plot.diamonds <- ggplot(diamonds, aes(clarity, fill = cut)) + geom_bar() +
  theme(axis.text.x = element_text(angle=70, vjust=0.5))
plot.diamonds

Now, arrange them on a grid

plot_grid(plot.mpg, plot.diamonds, labels = c("A", "B"))

More complex example

plot_grid(plot.mpg, plot.diamonds, plot.diamonds, plot.mpg, labels=c("A","B","C","D"),ncol=2,rel_heights = c(0.5,1))

Even more complex

• See here, which is part of the egg package.

Lattice + grid/gridExtra

library(lattice)
library(grid)
library(gridExtra)
lattice.mpg = xyplot(hwy~cty,data=mpg,group=factor(cyl),auto.key=T)
combined = grid.arrange(lattice.mpg, plot.mpg,ncol=2)
grid.draw(combined)

===================================================

lattice.diamonds = histogram(~clarity|cut,data=diamonds)
lattice.diamonds

Where layouts matter

• Talks, posters,grants
• No one will spend 10 minutes going over your figures, so boil them down to the basics!

Getting it done

• Filter out the data for panels from the original. dplyr
• Remove axis data from plots as appropriate
• Combine into one using grid.arrange, cowplot, etc.
• Generally, this involves "hacking" multiple existing plotting scripts into one.
• But this method is more reliable than Illustrator/Inkscape, and the new script documents exactly what you did!

Automated reports

• One method

Automated reports with a "driver" script

A driver can look like this:

# Define the input file name that the template expects
infile <- 'data2.txt'
# Execute the template
rmarkdown::render("ReportTemplate.Rmd")

(Show the template...)

littler

• littler will be your friend here.
• Pronounced "little r"
• Introduces proper scripting semantics to R OMG.

#!/usr/bin/env r
if(is.null(argv)) { stop("need some args") }
print("hi")
print(argv[1])

Annotated reports using Python

• Have your Python workflow generate any plots/tables and save them to files in a directory specified by the user
• Use knitr to render the report from an Rmd file that expects the directory name as a parameter.
• You include the graphics using regular Markdown syntax.

Annotated reports using Python

For Jupyter notebooks, you have:

• papermill
• Officially, the Jupyter project views this as "outside the scope"

More generally

• Read more about pandoc and let me know if you find out anything!

Plotting in Python

library(reticulate)
use_python("/usr/bin/python3")

Matplotlib

import matplotlib.pyplot as plt
import numpy as np
t = np.arange(0.,2.,0.01)
s = 1 + np.sin(2*np.pi*t)
plt.plot(t,s)
plt.grid(True)
plt.show()

The output

Seaborn

import seaborn as sns
tips = sns.load_dataset("tips")
g = sns.FacetGrid(tips, col="time")
g.map(plt.hist, "tip");
plt.show()

Example 2

g = sns.FacetGrid(tips, col="sex", hue="smoker")
g.map(plt.scatter, "total_bill", "tip", alpha=.7)
g.add_legend();
plt.show()

Python is GREAT for adding text!

for ax in g.axes.flat:
  ax.set_title(r'$\Theta = \frac{\Gamma}{\sum_{i=1}{10}}$')
plt.show()