MeetUpSlides.Rmd

---
title: "My Summer @ Rstudio & an intro to the scales pkg"
author: "Dana Seidel"
#date: "09/19/2018"
output:
  rmdshower::shower_presentation:
    self_contained: false
    katex: true
    ratio: 16x10
    theme: material
---


```{r setup, include=FALSE}
knitr::opts_chunk$set(
  collapse = TRUE, comment = "#>",
  fig.align = "center",
  fig.asp = 0.618, # 1 / phi
  fig.width = 5.5
)

library(scales)
library(tidyverse)
set.seed(4235)
```


## My Summer @ Rstudio & <br> an intro to the r-lib/scales `r emo::ji("package")`

Dana Paige Seidel 
<br>
R-Ladies MeetUp, September 19, 2018

<img src="TidyverseShower_files/scales.png">
<img src="TidyverseShower_files/ggplot2.png">
<img src="TidyverseShower_files/scales.png">
<img src="TidyverseShower_files/ggplot2.png">
<img src="TidyverseShower_files/scales.png">
<img src="TidyverseShower_files/ggplot2.png">

## My Story `r emo::ji("book")`

- Introduced to R in college biostats course in 2008
- BSc & MSc in biology/ecology
- Berkeley PhD Student studying animal movement `r emo::ji("elephant")` `r emo::ji("zebra")`
- Disillusioned with academia,<br>in `r emo::ji("heart")` with #Rstats, teaching, & data science
- 2016 Google Maps GeoDataAnalytics summer intern
- 2018 Rstudio ggplot2 summer intern
- Graduating in May 2019 `r emo::ji("celebrate")`

<img src="MeetUpSlides_files/me.jpg" class="place right">

# My Summer
## Three overarching goals

1. scales 1.0.0 release
2. implement new features and bug fixes in ggplot2 (3.0.0.9000)
3. document! document! document!

## Prepping scales 1.0.0
- moved into r-lib 
- issue triaged
- revived and ushered old PRs over the finish line
- updated readme & built [new pkgdown site](http://scales.r-lib.org){target="_blank"}
- Complete [release checklist](https://github.com/r-lib/scales/issues/157){target="_blank"}

## ggplot2 features, fixes, and docs

Since ggplot2 3.0.0 was release about halfway through my internship, 
I started with a lot of documentation. 

I made several PRs just doing careful review of documentation of the most visited reference
sites and general cleaning (spell-check, consistency)

Then I got into some features and fixes mostly regarding themes and secondary axes. 

## Keep an eye out for my changes `r emo::ji("magnifying_glass_tilted_right")`!

<strong>Coming soon to a ggplot2 near you...</strong> 

- Improvements to seconday axis functionality for transformed and date/datetime axes 
(this is in 3.0.1, out next month!)
- Themeable aesthetics -- setting unmapped aesthetics through themes. 
- Right to Left plotting

## Sneak Peek: Themeable aesthetics
```{r eval = F}
my_theme <- theme(geom = element_geom(colour = "purple", fill = "darkblue"))
ggplot(mpg, aes(displ, hwy)) + geom_point() + my_theme
```

<img src=TidyverseShower_files/WSQ6eod.png height=360 class="place bottom center">

## Bean counting
<font color="#96049b">scales 0.5.0.9000-1.0.0.9000</font>: authored 22 PRs, merged 40+ PRs total, 
24 contributors to the 1.0.0 release
  
<font color="#96049b">ggplot2, 3.0.0.9000</font>: opened 18 PRs (2 still open!)

Merged PRs in 3 tidyverse/r-lib packages: scales, ggplot2, and lubridate! 
Recently vdiffr too!

# Intro to Scales: Overview
## Scaling `r emo::ji("chart_with_upwards_trend")`

- converting data values to perceptual properties. 
- includes the inverse process: making guides (legends and axes) that can be used to read the graph 

Scaling and guides are often some of the most difficult parts of building any visualization.


## Scales `r emo::ji("package")`

The scales package provides the internal scaling infrastructure to
[ggplot2](github.com/tidyverse/ggplot2){target="_blank"} and exports standalone, system-agnostic, 
functions. 

<strong> Use scales to customize 
the transformations, breaks, guides and palettes in your visualizations.
</strong>

## Installation

```{r, eval = FALSE}
# Scales is installed when you install ggplot2 or the tidyverse.
# But you can install just scales from CRAN:
install.packages("scales")

# Or the development version from Github:
# install.packages("devtools")
devtools::install_github("r-lib/scales")

# let's load it too! Scales is imported by ggplot2 but not loaded explicitly
library(scales)

# For these slides, we'll also want
library(tidyverse)
# for dplyr and ggplot2!
```

# Palettes
## Colour palettes `r emo::ji("art")`

scales provides a number of color pallete functions that, given a range of values 
or the number of colours your want, will return a range of colors by hex code. 

```{r, palettes}
# pull a list of colours from any palette
viridis_pal()(4)
brewer_pal(type = "div", direction = -1)(4)
div_gradient_pal()(seq(0, 1, length.out = 4))
```

## Color palettes (cont.)
```{r}
# show_col is a quick way to view palette output
show_col(viridis_pal()(4))
```

## Use scales palettes with baseR
These functions are primarily used under the hood in ggplot2, but can be combined 
with any plotting system. For example, use them in combination with `grDevices::palette()`,
provided with base R, to affect your base plots...

## BaseR example
```{r fig.asp=.9, fig.width = 4.25}
palette(viridis_pal()(4))
plot(Sepal.Length ~ Sepal.Width, data = iris, col = Species, pch = 20)
```

## Non-color palettes
Often you want to be able to scale elements other than color. e.g. size, alpha, shape... 
Of course, scales handles those too!

```{r}
your_data <- runif(13, 1, 20)
area_pal(range = c(1, 20))(your_data)
shape_pal()(6)
```

## See these in action in ggplot2

```{r, eval=FALSE}
# color examples...
scale_fill_brewer()
scale_color_grey()
scale_color_viridis_c()

# shape examples
scale_shape()
scale_shape_ordinal()

# implement them yourself with...
scale_color_manual()
scale_shape_manual()
scale_size_manual()
# using available scales functions!
```

# Guides & breaks
## Making Guides easy: scales' formatters
The scales package also provides useful helper 
functions for formatting numeric data for all types of labels 

As of 1.0.0, most of scales formatters are just variations on the generic `number()`
and `number_format()` functions. 

## The number formatter
+ The generic formatter offers consistency in behaviour and arguments across formatters
and adds additional functionality for international customization.
+ With it's adoption, existing formatters gained new, consistent arguments for 
`scale`, `accuracy`, `trim`, `big.mark`, `decimal.mark`, `prefix`, `suffix` etc. 

## The number formatter, in action

By default, `number()` will take any numeric vector, round them to nearest whole
number, add spaces between every 3 digits and return a character vector useful for
feeding to a labels argument in ggplot2. 

```{r}
number(c(12.3, 4, 12345.789, 0.0002))
```

## Changing defaults

You can easily specify a different rounding behavior, or change the `big_mark` 
or `decimal_mark` for international 
styling. Even add a `prefix` or a `suffix` or `scale` your numbers on the fly. 

```{r}
number(c(12.3, 4, 12345.789, 0.0002),
  big.mark = ".",
  decimal.mark = ",",
  accuracy = .01
)
```

## Other Formatters:
<font size = "3">
`comma_format()` `comma()` `percent_format()` `percent()` `unit_format()`

`date_format()` `time_format()` : Formatted dates and times.

`dollar_format()` `dollar()` : Currency formatters, round to nearest cent and display dollar sign.

`ordinal_format()` `ordinal()` `ordinal_english()` `ordinal_french()` `ordinal_spanish()`: add ordinal suffixes (-st, -nd, -rd, -th) to numbers.

`pvalue_format()` `pvalue()` : p-values formatter

`scientific_format()` `scientific()` : Scientific formatter
</font>

## Examples 

```{r, formatters}
# percent() function takes a numeric and does your division and labelling for you
percent(c(0.1, 1 / 3, 0.56))

# comma() adds commas into large numbers for easier readability
comma(10e6)

# dollar() adds currency symbols
dollar(c(100, 125, 3000))

# unit_format() adds unique units
# the scale argument allows for simple conversion on the fly
unit_format(unit = "ha", scale = 1e-4)(c(10e6, 10e4, 8e3))
```

## Why the *_format() functions?

Where `number()` returns a character vector, `number_format()` and like functions
returns a fuctions that can be applied repeatedly or fed to a `labels` argument in a 
ggplot2 scale function. 

```{r}
# percent formatting in the French style
french_percent <- percent_format(decimal.mark = ",", suffix = " %")
french_percent(runif(10))

# currency formatting Euros (and simple conversion!)
usd_to_euro <- dollar_format(prefix = "", suffix = "\u20ac", scale = .86)
usd_to_euro(100)
```

## Applied in ggplot scales

```{r}
dsamp <- dplyr::sample_n(diamonds, 1000)
ggplot(dsamp, aes(x = carat, y = price, colour = clarity)) +
  geom_point() + scale_y_continuous(labels = usd_to_euro)
```

## Breaks
<font size="4">

`scales::extended_breaks()` sets most breaks by default in ggplot2

`pretty_breaks()` is an alternative break calculation 

Many of the formatter and transformation functions have matching break functions, eg:

- `log_breaks()` is used to set breaks for log transformed axes with `log_trans()`.
- `date_breaks()` is used to set nice breaks for date and date/time axes.

</font>

# Bounds & transformations
## Rescaling data
scales provides a handful of functions for rescaling data to fit new ranges.

```{r rescale}
# the rescale functions can rescale continuous vectors to new min, mid, or max values
x <- runif(5, 0, 1)
x

rescale(x, to = c(0, 50))
rescale_mid(x, mid = .25)
rescale_max(x, to = c(0, 50))
```

## Squish, Discard, Censor
```{r squish}
# squish() will squish your values into a specified range, respecting NAs
squish(c(-1, 0.5, 1, 2, NA), range = c(0, 1))

# discard will drop data outside a range, respecting NAs
scales::discard(c(-1, 0.5, 1, 2, NA), range = c(0, 1))

# censor will return NAs for values outside a range
censor(c(-1, 0.5, 1, 2, NA), range = c(0, 1))
```

## Applied to ggplot2
<font size= "4">Squish can be really useful for setting the `oob` argument for a colour scale with reduced limits.</font>

```{r fig.width=4.25}
ggplot(iris, aes(x = Sepal.Length, y = Sepal.Width, colour = Sepal.Length)) +
  geom_point() + scale_color_continuous(limit = c(6, 8), oob = scales::squish)
```

## Transformations
scales provides a number of common transformation functions (`*_trans()`) which 
specify functions to preform data transformations, format labels, and set correct breaks.

For example:
`log_trans()`, `sqrt_trans()`, `reverse_trans()` power the `scale_*_log10()`,
`scale_*_sqrt()`, `scale_*_reverse()` functions in ggplot2. 

## Additional Transformations
<font size="4">

- `asn_trans()`: Arc-sin square root transformation.
- `atanh_trans()`: Arc-tangent transformation.
- `boxcox_trans()` `modulus_trans()` : Box-Cox & modulus transformations.
- `date_trans()` `time_trans()` `hms_trans()` : transformations for date, datetime, and hms classes
- `exp_trans()` : Exponential transformation (inverse of log transformation).
- `pseudolog_trans()` : Pseudo-log transformation
- `probabilty_trans()`: Probability transformation

and more... 
</font>

## Building your own transformations

scales also gives users the ability to define and apply their own custom 
transformation functions for repeated use.

```{r transforms}
# use trans_new to build a new transformation
dollar_log <- trans_new(
  name = "dollar_log",
  trans = log_trans(base = 10)$trans, # extract a single element from another trans
  inverse = function(x) 10^(x), # or write your own custom functions
  breaks = log_breaks(),
  format = dollar_format()
  )
```

## Applied in ggplot2
```{r}
# apply our new transformation!
ggplot(dsamp, aes(x = carat, y = price, colour = clarity)) +
  geom_point() + scale_y_continuous(trans = dollar_log)
```
  
## Additional uses

In 1.0.0.9000, scales implements `Range()` functions to allow users to create their own
scales and mutable ranges. These were exported in 1.0.0 but had fatal bugs now fixed in the dev version.

These functions will eventually be imported into ggplot2 to power custom ranges
instead of ggproto objects. 

# Wrap Up!
## Takeaway
<font color="#96049b">scales</font> is a useful package for specifying breaks, labels, 
palettes, and transformations for your visualizations in ggplot2 and beyond. 

## Summer Revelations 
Open source development is just that open! Open to me AND to you! <strong>We need more ladies in dev!</strong>

Development work is a wonderful blend of creativity, investigation, puzzle solving, and
design. In my view, the pefect hobby and a unique way to give back to the #rstats community.

I want to use my experience in any way I can to help other women get
involve in their favorite packages or creating their own!

## Summer Revelations (cont.)
Want to know more about what I did this summer? Read my [blog](https://www.danaseidel.com/2018-09-01-ATidySummer/){target="_blank"} 
about the experience and my work. 

## Questions?

Slides available at [danaseidel.com/MeetUpSlides](https://www.danaseidel.com/MeetUpSlides){target="_blank"}

`r emo::ji("package")` [rmdshower](https://github.com/MangoTheCat/rmdshower){target="_blank"}

`r emo::ji("package")` [emo](https://github.com/hadley/emo){target="_blank"}

For the raw .Rmd for these slides, see [here](https://github.com/dpseidel/dpseidel.github.io/blob/master/MeetUpSlides.Rmd){target="_blank"}.

For the adapted css code for this <font color="#96049b">#Rladies</font> theme, see [here](https://github.com/dpseidel/dpseidel.github.io/blob/master/MeetUpSlides_files/shower-material/package/styles/screen-16x10.css){target="_blank"}.