In andrewGhazi/geryon: simple utility functions

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

geryon

This is a package containing small utility functions.

Installation

You can install the released version of geryon from this Github repo.

remotes::install_github("andrewGhazi/geryon")

Functionality

Right now this package contains only a handful of functions: ws_size, pull1, get/add_local_density and theme_pres.

ws_size

It's easy to see the memory usage of a single object with object.size but doing that in a sorted, pretty way for everything in the workspace is a bit more involved. This function has taken care of that.

> a = 1:10
> tmp = mtcars
> ws_size()
# A tibble: 2 x 3
    obj obj_size n_bytes
  <chr>    <chr>   <int>
1   tmp   6.6 Kb    6736
2     a 88 bytes      88

pull1

dplyr::pull is a convenient way to pull out the values of a column from a data frame, but if you've got a long list column and just want to check one to make sure everything's working, printing the whole list can clutter up the console. pull1 just pulls out one.

> tmp = data_frame(mu = 5:10, samples = purrr::map(mu, ~rnorm(10, mean = .x, sd = 1)))
> tmp
# A tibble: 6 x 2
     mu    samples
  <int>     <list>
1     5 <dbl [10]>
2     6 <dbl [10]>
3     7 <dbl [10]>
4     8 <dbl [10]>
5     9 <dbl [10]>
6    10 <dbl [10]>
> tmp %>% pull1(samples)
 [1] 4.651063 3.986187 5.421502 4.554197 4.880644 4.511554 5.572671 4.948658 6.132115 5.545000

theme_pres

This is simply a modified version of ggplot2::theme_light with larger text and dark facet labels. This makes it easier to prepare easy-to-read plots for presentations.

diamonds %>% sample_n(1000) %>% ggplot(aes(carat, price)) + geom_point() + theme_pres() + facet_wrap('cut')

get/add_local_density

This is handy for dealing with severe overplotting in situations where you still want to plot individual points rather than a density estimate. You simply give add_local_density the two variables used in your plot, then you aesthetically map the resulting local_density column to color.

library(ggplot2)
library(magrittr)
overplotted = data.frame(x = rt(1e4, df = 5),
                         y = rt(1e4, df = 5))
overplotted %>% 
  ggplot(aes(x,y)) +
  geom_point(alpha = .1) + 
  theme_light() + 
  labs(title = expression('Intractable overplotting, even with'~alpha~' = 0.1'))

knitr::include_graphics('man/figures/overplotted.png')

library(geryon)
overplotted %>% 
  add_local_density(x, y) %>% 
  ggplot(aes(x,y)) +
  geom_point(aes(color = local_density)) + # <-- this is the important bit
  theme_light() + 
  scale_color_viridis_c() + 
  labs(title = 'Coloring by local density naturally transitions between\nscattered points to a density estimate')

knitr::include_graphics('man/figures/color_density.png')

This is a trivial example but sometimes there's simultaneously important structure hidden in the overplotted region AND important meaning associated with individual points in the sparse outer regions. This sort of plot shows both.

I'm working on a color_density_scatter function that goes straight from the data to the plot.

`insert_img_link()`

This function inserts an image link (![](images/blah.png)) to an image on your clipboard into the current source document at the cursor. This way you don't have to fuss around with the file manager to get images into your Quarto report.

It's provided as an RStudio addin so you can bind a keyboard shortcut to it. I use Ctrl+Shift+V.

This function makes some assumptions and currently doesn't do many checks. There are some caveats:

It only inserts PNGs. I'll probably add jpg sooner or later.
It won't work if you currently have multiple images on your clipboard.
It uses xclip to copy from the image from the clipboard to the images/ directory. That must be installed.
There must be an images/ directory in the same directory as the source document.
It doesn't clean up after itself. If you use it a lot, you might end up with many copies of the same image in your images directory.