Seamless data-driven reporting with {epoxy}

class: bottom, left, title-slide

.title[
# Seamless data-driven reporting with {epoxy}
]
.author[
### Garrick Aden-Buie
]
.date[
### R/Medicine: 2023-06-08
]

---

background-size: cover
background-position: bottom right
background-image: url(images/tanaphong-toochinda-GagC07wVvck-unsplash.jpg)

<STYLE type='text/css' scoped>
PRE.fansi SPAN {padding-top: .25em; padding-bottom: .25em};
</STYLE>

.footnote[
.bg-white-blur[Photo by <a href="https://unsplash.com/photos/GagC07wVvck">Tanaphong Toochinda</a>]
]

???
I recently learned an interesting fact about childcare costs in the U.S.
in a report by the Department of Labor:

---
layout: true
name: quote

.fancy-quote.f3.lh-copy.mt5[
A study of childcare prices in
.hl.number[2,360]
U.S. counties across
.hl.number[47]
states shows that childcare prices
for a single child
range from
.hl.number[$4,810]
.hl.qualifier[for school-age home-based care in small counties]
to .hl.number[$15,417]
.hl.qualifier[for infant center-based care in very large counties].
This is equivalent to between
.hl.number[8%] and .hl.number[19.3%]
of median family income per child in paid care.
]

.footnote[
[National Database of Childcare Prices](https://www.dol.gov/newsroom/releases/wb/wb20230124)
]

---

???
\[read quote\]

If you're anything like me, one part of you is saying "wow, children are expensive".
(at least in the U.S.)
The other part of you is thinking about the raw data behind this sentence.

You're probably noticing the qualifiers...

---
class: hl-qualifier

???

You're probably noticing the qualifiers in this quote,
like [read qualifiers].

You're thinking about how these qualifiers map
to decisions about how the data was summarized.

And you're definitely thinking about all of the numbers in this quote...

---
class: hl-number

???
... like [read numbers]

And now you're completely distracted thinking about the process behind this quote.

* Where did the data come from?
* How hard was it to clean?
* How did it go from rows in a table to this sentence in the report?
* I mean that literally, like, what happens if we find an error or a state updates a CSV file somewhere? Who's job is it to come back to this sentence and update it?

---
layout: false
background-image: url(images/la-rel-easter-KuCGlBXjH_o-unsplash.jpg)
background-size: 115%
background-position: bottom center

.footnote.white[
Photo by <a href="https://unsplash.com/photos/KuCGlBXjH_o">La-Rel Easter</a>
]

???

When we talk about data-driven reporting,
I think it's important to acknowledge how many decisions
we pack into the tight spaces between our words.

---

```r
childcare_summary
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 32 × 5
#&gt; county_size type age cost mhi_pct
#&gt; &lt;fct&gt; &lt;chr&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt;
#&gt; 1 small center school-age 5288. 0.109 
#&gt; 2 small home school-age 4810 0.0991
#&gt; 3 small center infant 7461. 0.154 
#&gt; 4 small center toddler 6760 0.139 
#&gt; 5 small center preschool 6239. 0.129 
#&gt; 6 small home infant 5824 0.120 
#&gt; 7 small home toddler 5713. 0.118 
#&gt; 8 small home preschool 5541. 0.114 
#&gt; 9 mid-sized center school-age 6732. 0.115 
#&gt; 10 mid-sized home school-age 6360. 0.108 
#&gt; # ℹ 22 more rows
</CODE></PRE>

???

All of these descisions manifest in a data frame or two.
And it's extra satisfying when they're tidy.

This table plugs in very nicely into dense tables and plots,
which we're as data scientists lean on heavily.

---
background-image: url('images/map-plot.png')
background-size: 85%
background-position: bottom center

.clip[
Slide image: a map of the U.S. with each county shaded by median yearly price of
infant center-based care.
Chester County, PA is highlighted and an pop-over lists related statistics about the county.
]

.footnote[
.f6[[Women's Bureau](https://www.dol.gov/agencies/wb/topics/childcare/price-by-age-care-setting)]
]

???

Like this map of the U.S.
comparing the median yearly price of infant center-based care.

This map is great (and it's even interactive!)
but it has its place.
No reasonable person is going to use this map
and arrive at the conclusions of our quote.

---

???

Or this super basic horizontal bar plot.
I just took the table and passed it to ggplot,
mapping childcare costs to the x-axis and the child's age group to the y-axis.
I faceted by county size and I set the bar color to the type of care:
home-based or center-based care.

---

```r
ggplot(childcare_summary) +
  aes(cost, age, fill = type) +
  geom_col(position = "dodge") +
  facet_wrap(~ county_size) +
  theme_grey(14)
```

???

This only takes a few lines of code.

But ggplot also gives you all kinds of tools
to help you take the raw data and map it
to labels and text that's easier to understand.

---

```r
ggplot(childcare_summary) +
  aes(cost, age, fill = type) +
  geom_col(position = "dodge") +
  facet_wrap(
    ~ county_size,
    labeller = labeller(county_size = \(x) str_to_title(paste(x, "County")))
  ) +
  labs(
    x = "Median yearly cost",
    y = "Age group",
    fill = "Type of care"
  ) +
  scale_fill_discrete(labels = \(x) paste0(x, "-based")) +
  scale_x_continuous(
    labels = scales::dollar_format(scale_cut = scales::cut_short_scale())
  ) +
  theme_gray(14)
```

???

With a few more lines of code

---

```r
ggplot(childcare_summary) +
  aes(cost, age, fill = type) +
  geom_col(position = "dodge") +
  facet_wrap(
    ~ county_size,
    labeller = labeller(county_size = \(x) str_to_title(paste(x, "County")))
  ) +
* labs(
*   x = "Median yearly cost",
*   y = "Age group",
*   fill = "Type of care"
* ) +
  scale_fill_discrete(labels = \(x) paste0(x, "-based")) +
  scale_x_continuous(
    labels = scales::dollar_format(scale_cut = scales::cut_short_scale())
  ) +
  theme_gray(14)
```

???

we can give the axis and legend appropriate labels

---

```r
ggplot(childcare_summary) +
  aes(cost, age, fill = type) +
  geom_col(position = "dodge") +
  facet_wrap(
    ~ county_size,
    labeller = labeller(county_size = \(x) str_to_title(paste(x, "County")))
  ) +
  labs(
    x = "Median yearly cost",
    y = "Age group",
    fill = "Type of care"
  ) +
  scale_fill_discrete(labels = \(x) paste0(x, "-based")) +
* scale_x_continuous(
*   labels = scales::dollar_format(scale_cut = scales::cut_short_scale())
* ) +
  theme_gray(14)
```

???

And we can format the x-axis in U.S. dollars

---

```r
ggplot(childcare_summary) +
  aes(cost, age, fill = type) +
  geom_col(position = "dodge") +
  facet_wrap(
    ~ county_size,
*   labeller = labeller(county_size = \(x) str_to_title(paste(x, "County")))
  ) +
  labs(
    x = "Median yearly cost",
    y = "Age group",
    fill = "Type of care"
  ) +
* scale_fill_discrete(labels = \(x) paste0(x, "-based")) +
  scale_x_continuous(
    labels = scales::dollar_format(scale_cut = scales::cut_short_scale())
  ) +
  theme_gray(14)
```

???

or give the facets and legends better labels

To end up with a plot like this...

---

???

But these updates all happen between our data frame
and the last few lines of code.

---

```r
childcare_summary
```

???

Meaning that we didn't have to change our source data frame at all.

---
class: middle f1 highlight-last-item

<h2 class="f1" style="font-size:3rem">epoxy</h2>

* .f3[Handle .pink.b[last-mile] transformations]

???

I started working on epoxy because I wanted
better tools that help you make these kinds of
last-mile transformations easier.

* .f3[Seamlessly use data in .pink.b[the prose] of reports and apps]

???

Tools that make it seamless to blend data into prose

* .f3[Re-usability through .pink.b[templates] and global settings]

???

And that improve usability through templates and global settings

---
class: middle center animated slideInDown slideOutDown

.center.w-50[
![glue hex sticker](images/glue.svg)
]

???

The good news is that there's an awesome package called glue
that does most of what I wanted but for strings in R.

It's biggest limitation is that it's a developer-oriented package,
meaning that it's designed to do one thing well with a minimal API
and basically zero dependencies.

---
class: middle center animated slideInDown

.center.w-50[
![epoxy hex sticker](images/epoxy.svg)
]

???

So you could consider epoxy an extension package for glue,
which also explains the name and the hex sticker.

---
class: middle

.f2.code[
install.packages(.pink["epoxy"]) 
.silver[pak::pak("gadenbuie/epoxy")] 
 
library(.pink[epoxy]) 
.silver[library(glue)]
]

???

You can install epoxy from CRAN and load it in the usual ways.

I'm loading glue here too for later in the presentation,
but you don't need to do that to use epoxy.

---
class: inverse center middle

# Let's try this together

.f3.f-zilla.dib.animated.lightSpeedIn.delay-1s[(speed run)]

---
layout: true

## TidyTuesday (2023, Week 19)

---

```r
library(tidyverse)

tuesdata <- tidytuesdayR::tt_load(2023, week = 19)

childcare_costs <- tuesdata$childcare_costs
counties <- tuesdata$counties
```

???

First of all, huge thanks to the TidyTuesday project where I discovered this dataset.
They also make it super easy to download the dataset and start working with it.

---

```r
counties
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 3,144 × 4
#&gt; county_fips_code county_name state_name state_abbreviation
#&gt; &lt;dbl&gt; &lt;chr&gt; &lt;chr&gt; &lt;chr&gt; 
#&gt; 1 1001 Autauga County Alabama AL 
#&gt; 2 1003 Baldwin County Alabama AL 
#&gt; 3 1005 Barbour County Alabama AL 
#&gt; 4 1007 Bibb County Alabama AL 
#&gt; 5 1009 Blount County Alabama AL 
#&gt; 6 1011 Bullock County Alabama AL 
#&gt; 7 1013 Butler County Alabama AL 
#&gt; 8 1015 Calhoun County Alabama AL 
#&gt; 9 1017 Chambers County Alabama AL 
#&gt; 10 1019 Cherokee County Alabama AL 
#&gt; # ℹ 3,134 more rows
</CODE></PRE>

???

The data come with two data sets (which were originally one set).
Counties and states...

---

```r
childcare_costs
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 34,567 × 61
#&gt; county_fips_code study_year unr_16 funr_16 munr_16 unr_20to64 funr_20to64
#&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt;
#&gt; 1 1001 2008 5.42 4.41 6.32 4.6 3.5
#&gt; 2 1001 2009 5.93 5.72 6.11 4.8 4.6
#&gt; 3 1001 2010 6.21 5.57 6.78 5.1 4.6
#&gt; 4 1001 2011 7.55 8.13 7.03 6.2 6.3
#&gt; 5 1001 2012 8.6 8.88 8.29 6.7 6.4
#&gt; 6 1001 2013 9.39 10.3 8.56 7.3 7.6
#&gt; 7 1001 2014 8.5 9.18 7.95 6.8 6.8
#&gt; 8 1001 2015 7.64 8.45 6.91 5.9 6.1
#&gt; 9 1001 2016 5.59 6.27 4.99 4.4 4.6
#&gt; 10 1001 2017 5.21 5.84 4.64 4.6 4.9
#&gt; # ℹ 34,557 more rows
#&gt; # ℹ 54 more variables: munr_20to64 &lt;dbl&gt;, flfpr_20to64 &lt;dbl&gt;,
#&gt; # flfpr_20to64_under6 &lt;dbl&gt;, flfpr_20to64_6to17 &lt;dbl&gt;,
#&gt; # flfpr_20to64_under6_6to17 &lt;dbl&gt;, mlfpr_20to64 &lt;dbl&gt;, pr_f &lt;dbl&gt;,
#&gt; # pr_p &lt;dbl&gt;, mhi_2018 &lt;dbl&gt;, me_2018 &lt;dbl&gt;, fme_2018 &lt;dbl&gt;, mme_2018 &lt;dbl&gt;,
#&gt; # total_pop &lt;dbl&gt;, one_race &lt;dbl&gt;, one_race_w &lt;dbl&gt;, one_race_b &lt;dbl&gt;,
#&gt; # one_race_i &lt;dbl&gt;, one_race_a &lt;dbl&gt;, one_race_h &lt;dbl&gt;, …
</CODE></PRE>

???

And the main attraction, the child care costs.

---
layout: false

## Data Dictionary: `childcare_costs`

---

.flex[
.w-25[
## Prepare

* 2018 only

* Where reported

* Pick: population, income, price

* County size
]

.w-75.pl2[

```r
childcare_summary <-
 childcare_costs |>
 filter(study_year == 2018, !is.na(mc_infant)) |>
 select(1:2, total_pop, mhi_2018, matches("mfcc|mc")) |>
 mutate(
 county_size = case_when(
 total_pop < 1e5 ~ "small",
 total_pop < 5e5 ~ "mid-sized",
 total_pop < 1e6 ~ "large",
 TRUE ~ "very large"
 ),
 county_size = fct_reorder(county_size, total_pop)
 )
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 2,360 × 13
#&gt; county_fips_code study_year total_pop mhi_2018 mcsa
#&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt;
#&gt; 1 1001 2018 55200 58786 95.9
#&gt; 2 1003 2018 208107 55962 108. 
#&gt; 3 1005 2018 25782 34186 79.2
#&gt; 4 1007 2018 22527 45340 102. 
#&gt; 5 1009 2018 57645 48695 127. 
#&gt; # ℹ 2,355 more rows
#&gt; # ℹ 8 more variables: mfccsa &lt;dbl&gt;, mc_infant &lt;dbl&gt;,
#&gt; # mc_toddler &lt;dbl&gt;, mc_preschool &lt;dbl&gt;,
#&gt; # mfcc_infant &lt;dbl&gt;, mfcc_toddler &lt;dbl&gt;,
#&gt; # mfcc_preschool &lt;dbl&gt;, county_size &lt;fct&gt;
</CODE></PRE>
]
]

---

.flex[
.w-25[
## Pivot

* Collect all costs

* Extract information from variable names

* Recode age and type

* Convert to yearly cost
]

.w-75.pl2[

```r
childcare_summary <-
 childcare_summary |>
 pivot_longer(
 cols = matches("mfcc|mc"),
 names_to = c("type", "age"),
 names_pattern = "(mfcc|mc)(.*)",
 values_to = "cost"
 ) |>
 mutate(
 age = str_remove(age, "^_"),
 age = recode(age, sa = "school-age"),
 age = factor(age, c("infant", "toddler", "preschool", "school-age")),
 type = recode(type, mfcc = "home", mc = "center"),
 cost = cost * 52
 )
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 18,880 × 8
#&gt; county_fips_code study_year total_pop mhi_2018 county_size
#&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;dbl&gt; &lt;fct&gt; 
#&gt; 1 1001 2018 55200 58786 small 
#&gt; 2 1001 2018 55200 58786 small 
#&gt; 3 1001 2018 55200 58786 small 
#&gt; 4 1001 2018 55200 58786 small 
#&gt; # ℹ 18,876 more rows
#&gt; # ℹ 3 more variables: type &lt;chr&gt;, age &lt;fct&gt;, cost &lt;dbl&gt;
</CODE></PRE>
]
]

---

.flex[
.w-25[
## Summarize

* Median cost and income

* By county size, care type and age group

* Cost as a % of income
]

.w-75.pl2[

```r
childcare_summary <-
 childcare_summary |>
 summarize(
 across(cost, \(x) median(x, na.rm = TRUE)),
 across(mhi_2018, \(x) median(x, na.rm = TRUE)),
 .by = c(county_size, type, age)
 ) |>
 mutate(mhi_pct = cost / mhi_2018) |>
 select(-mhi_2018)
```

---

.flex[
.w-25[
## Quote

* Min/max cost
]

.w-75.pl2[

```r
costs_min <-
 childcare_summary |>
 filter(mhi_pct == min(mhi_pct))
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 1 × 5
#&gt; county_size type age cost mhi_pct
#&gt; &lt;fct&gt; &lt;chr&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt;
#&gt; 1 small home school-age 4810 0.0991
</CODE></PRE>

```r
costs_max <-
 childcare_summary |>
 filter(mhi_pct == max(mhi_pct))
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 1 × 5
#&gt; county_size type age cost mhi_pct
#&gt; &lt;fct&gt; &lt;chr&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt;
#&gt; 1 very large center infant 15417. 0.238
</CODE></PRE>
]
]

---

.flex[
.w-25[
## Quote

* Also counts of counties and states
]

.w-75.pl2[

```r
counties_states <-
 childcare_costs |>
 filter(study_year == 2018, !is.na(mc_infant)) |>
 left_join(counties, by = "county_fips_code") |>
 distinct(state_abbreviation, county_name)

n_counties <- nrow(counties_states)
```

```
#> [1] 2360
```

```r
n_states <- n_distinct(counties_states$state_abbreviation)
```

```
#> [1] 41
```
]
]

---
class: inverse center middle

# How do we get this into prose?

---

```r
costs_min
```

```r
costs_max
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 1 × 5
#&gt; county_size type age cost mhi_pct
#&gt; &lt;fct&gt; &lt;chr&gt; &lt;fct&gt; &lt;dbl&gt; &lt;dbl&gt;
#&gt; 1 very large center infant 15417. 0.238
</CODE></PRE>

---
layout: true

## R Markdown is amazing

---

```markdown
A study of childcare prices in
2,360
U.S. counties across
47
states shows that childcare prices
for a single child
range from
$4,810
for school-age home-based care
in small counties to
$15,417
for infant center-based care
in very large counties.
This is equivalent to between
8% and 19.3%
of median family income per child in paid care.
```

???

Here's our sentence as if we've written it out in an `.Rmd` file.

I've broken it up into a lot of lines so that the changes are easier to read as we modify this.

---

```markdown
A study of childcare prices in
____
U.S. counties across
____
states shows that childcare prices
for a single child
range from
$____
for school-age home-based care
in small counties to
$____
for infant center-based care
in very large counties.
This is equivalent to between
____% and ____%
of median family income per child in paid care.
```

???

But when we're coming up with a sentence like this, we don't know what the numbers are going to be.

So let's replace these with blanks.

---

```markdown
A study of childcare prices in
____
U.S. counties across
____
states shows that childcare prices
for a single child
range from
$____
for school-age home-based care
in ____ counties to
$____
for infant center-based care
in ____ counties.
This is equivalent to between
____% and ____%
of median family income per child in paid care.
```

???

We also don't know which county sizes we're going to be talking about.

---

```markdown
A study of childcare prices in
____
U.S. counties across
____
states shows that childcare prices
for a single child
range from
$____
for school-age ____-based care
in ____ counties to
$____
for infant ____-based care
in ____ counties.
This is equivalent to between
____% and ____%
of median family income per child in paid care.
```

???

Or the type of childcare or educational center

---

```markdown
A study of childcare prices in
____
U.S. counties across
____
states shows that childcare prices
for a single child
range from
$____
for ____ ____-based care
in ____ counties to
$____
for ____ ____-based care
in ____ counties.
This is equivalent to between
____% and ____%
of median family income per child in paid care.
```

???

Or even which age group will be involved in the least and most expensive categories.

---

```markdown
A study of childcare prices in
*`r n_counties`
U.S. counties across
*`r n_states`
states shows that childcare prices
for a single child
range from
*$____
*for ____ ____-based care
*in ____ counties to
*$____
*for ____ ____-based care
*in ____ counties.
This is equivalent to between
*____% and ____%
of median family income per child in paid care.
```

???

And then we work our way forward, using inline code chunks to blend our data into the text.

---

```markdown
A study of childcare prices in
*`r n_counties`
U.S. counties across
*`r n_states`
states shows that childcare prices
for a single child
range from
*`r costs_min$cost`
*for `r costs_min$age` `r costs_min$type`-based care
*in `r costs_min$county_size` counties to
*$____
*for ____ ____-based care
*in ____ counties.
This is equivalent to between
*`r costs_min$mhi_pct` and ____%
of median family income per child in paid care.
```

???

Then all the pieces we know from the minimum cost category

---

```markdown
A study of childcare prices in
*`r n_counties`
U.S. counties across
*`r n_states`
states shows that childcare prices
for a single child
range from
*`r costs_min$cost`
*for `r costs_min$age` `r costs_min$type`-based care
*in `r costs_min$county_size` counties to
*`r costs_max$cost`
*for `r costs_max$age` `r costs_max$type`-based care
*in `r costs_max$county_size` counties.
This is equivalent to between
*`r costs_min$mhi_pct` and `r costs_max$mhi_pct`
of median family income per child in paid care.
```

???

and everything we know from the maximum cost category

---
layout: true

## R Markdown is amazing.pink[-ish]

---

A study of childcare prices in 
2360 
U.S. counties across 
41 
states shows that childcare prices 
for a single child 
range from 
4810 
for school-age home-based care 
in small counties to 
1.541696\times 10^{4} 
for infant center-based care 
in very large counties. 
This is equivalent to between 
0.0990792 and 0.2376447 
of median family income per child in paid care.

???

And then we render it... and it's amazing.
I mean, it's fine.
It's great that we're not copying and pasting numbers into our documents!
But this isn't quite right.

---

```markdown
A study of childcare prices in
*`r format(n_counties, big.mark = ",")`
U.S. counties across
*`r format(n_states, big.mark = ",")`
states shows that childcare prices
for a single child
range from
*$`r format(costs_min$cost, big.mark = ",")`
*for `r costs_min$age` `r costs_min$type`-based care
*in `r costs_min$county_size` counties
*$`r format(round(costs_max$cost, 0), big.mark = ",")`
*for `r costs_max$age` `r costs_max$type`-based care
*in `r costs_max$county_size` counties.
This is equivalent to between
*`r round(costs_min$mhi_pct * 100, 2)`% and `r round(costs_max$mhi_pct * 100, 2)`%
of median family income per child in paid care.
```

???

But these are inline code chunks, right, so we can just write some code around the values.

---
layout: true
class: center middle

## Spicy hot-take time

---

.f2.absolute.center.left-0.w-100[
.animated.fadeInDownBig.dib.delay-staggered[&#x1F336;&#xFE0F;]
.animated.fadeInDownBig.dib.delay-staggered[.o-30[&#x1F336;&#xFE0F;]]
.animated.fadeInDownBig.dib.delay-staggered[.o-30[&#x1F336;&#xFE0F;]]
.animated.fadeInDownBig.dib.delay-staggered[.o-30[&#x1F336;&#xFE0F;]]
.animated.fadeInDownBig.dib.delay-staggered[.o-30[&#x1F336;&#xFE0F;]]
]

---

.f2.absolute.center.left-0.w-100.f-pen.animated.slideInLeft[
Write R code where it can be R code.
]

---
layout: false
class: middle

### .red[Bad]

```markdown
prices range from $`r format(round(costs_max$cost, 0), big.mark = ",")`
```

### .silver[Better]

````markdown
```{r}
costs_max$cost <- format(round(costs_max$cost, 0), big.mark = ",")
```

prices range from $`r costs_max$cost`
````

### .green[Better-er]

````markdown
```{r}
costs_max$cost <- scales::dollar(costs_max$cost, accuracy = 1)
```

prices range from `r costs_max$cost`
````

---
class: center middle

.f3.absolute.center.left-0.w-100.f-pen.animated.slideInLeft[
Inline code is for the simplist code you cannot possibly mess up.
]

---
layout: true
class: animated fadeIn

## R Markdown .pink[&plus; epoxy]

---

```markdown
A study of childcare prices in
`r n_counties`
U.S. counties across
`r n_states`
states shows that childcare prices
for a single child
range from
`r costs_min$cost`
for `r costs_min$age` `r costs_min$type`-based care
in `r costs_min$county_size` counties to
`r costs_max$cost`
for `r costs_max$age` `r costs_max$type`-based care
in `r costs_max$county_size` counties.
This is equivalent to between
`r costs_min$mhi_pct` and `r costs_max$mhi_pct`
of median family income per child in paid care.
```

---

````markdown
*```{r}
*library(epoxy)
*```

A study of childcare prices in
`r n_counties`
U.S. counties across
`r n_states`
states shows that childcare prices
for a single child
range from
`r costs_min$cost`
for `r costs_min$age` `r costs_min$type`-based care
in `r costs_min$county_size` counties to
`r costs_max$cost`
for `r costs_max$age` `r costs_max$type`-based care
in `r costs_max$county_size` counties.
This is equivalent to between
`r costs_min$mhi_pct` and `r costs_max$mhi_pct`
of median family income per child in paid care.
````

---

````markdown
```{r}
library(epoxy)
```

*```{epoxy}
A study of childcare prices in
`r n_counties`
U.S. counties across
`r n_states`
states shows that childcare prices
for a single child
range from
`r costs_min$cost`
for `r costs_min$age` `r costs_min$type`-based care
in `r costs_min$county_size` counties to
`r costs_max$cost`
for `r costs_max$age` `r costs_max$type`-based care
in `r costs_max$county_size` counties.
This is equivalent to between
`r costs_min$mhi_pct` and `r costs_max$mhi_pct`
of median family income per child in paid care.
*```
````

---

````markdown
```{r}
library(epoxy)
```

```{epoxy}
A study of childcare prices in
*`r n_counties`
U.S. counties across
*`r n_states`
states shows that childcare prices
for a single child
range from
*`r costs_min$cost`
*for `r costs_min$age` `r costs_min$type`-based care
*in `r costs_min$county_size` counties to
*`r costs_max$cost`
*for `r costs_max$age` `r costs_max$type`-based care
*in `r costs_max$county_size` counties.
This is equivalent to between
*`r costs_min$mhi_pct` and `r costs_max$mhi_pct`
of median family income per child in paid care.
```
````

---

````markdown
```{r}
library(epoxy)
```

```{epoxy}
A study of childcare prices in
*{n_counties}
U.S. counties across
*{n_states}
states shows that childcare prices
for a single child
range from
*{costs_min$cost}
*for {costs_min$age} {costs_min$type}-based care
*in {costs_min$county_size} counties to
*{costs_max$cost}
*for {costs_max$age} {costs_max$type}-based care
*in {costs_max$county_size} counties.
This is equivalent to between
*{costs_min$mhi_pct} and {costs_max$mhi_pct}
of median family income per child in paid care.
```
````

---

````markdown
```{r}
library(epoxy)
```

```{epoxy}
A study of childcare prices in
{n_counties}
U.S. counties across
{n_states}
states shows that childcare prices
for a single child
range from
{costs_min$cost}
for {costs_min$age} {costs_min$type}-based care
in {costs_min$county_size} counties to
{costs_max$cost}
for {costs_max$age} {costs_max$type}-based care
in {costs_max$county_size} counties.
This is equivalent to between
{costs_min$mhi_pct} and {costs_max$mhi_pct}
of median family income per child in paid care.
```
````

---
layout: false
class: center middle

---
layout: false

## R Markdown .pink[&plus; epoxy] is amazing.pink.animated.dib.rotateInDownLeft.delay-2s[-ish]

A study of childcare prices in 
2360 
U.S. counties across 
41 
states shows that childcare prices 
for a single child 
range from 
4810 
for school-age home-based care 
in small counties to 
15416.96 
for infant center-based care 
in very large counties. 
This is equivalent to between 
0.0990792427956413 and 0.237644665043006 
of median family income per child in paid care.

---
layout: true
class: center middle
---

.code.f1[
&#96;r costs_max$.primary[cost]&#96;
]

---

.code.f1[
{costs_max$.primary[cost]}
]

---

.code.f1[
{.pink[.dollar] costs_max$.primary[cost]}
]

---

.code.f1[
{.pink[.pct] costs_max$.primary[mhi_pct]}
]

---

.code.f1[
{.pink[.comma] .primary[n_counties]}
]

.code.f3[
scales::label_.pink[comma]()(.primary[n_counties])
]

---
layout: true
class: animated fadeIn

## R Markdown .pink[&plus; epoxy] is amazing.pink.animated.dib.jackInTheBox[-er]

---

````markdown
```{r include=FALSE}
library(epoxy)
```

```{epoxy}
A study of childcare prices in {.comma n_counties} U.S. counties across
{n_states} states shows that childcare prices for a single child range from
{.dollar costs_min$cost} for {costs_min$age} {costs_min$type}-based care
in {costs_min$county_size} counties to {.dollar costs_max$cost} for
{costs_max$age} {costs_max$type}-based care in {costs_max$county_size} counties.
This is equivalent to between {.pct costs_min$mhi_pct} and
{.pct costs_max$mhi_pct} of median family income per child in paid care.
```
````

---

A study of childcare prices in 2,360 U.S. counties across
41 states shows that childcare prices for a single child range from
$4,810 for school-age home-based care
in small counties to $15,416.96 for
infant center-based care in very large counties.
This is equivalent to between 10% and
24% of median family income per child in paid care.

---
layout: true

## .pink[epoxy] is built on .orange[glue]

---

```r
glue("for a single child range from {costs_min$cost}")
```

```
#> for a single child range from 4810
```

```r
epoxy("for a single child range from {costs_min$cost}")
```

```
#> for a single child range from 4810
```

---

```r
glue("for a single child range from {.dollar costs_min$cost}")
```

```
#> Error in parse(text = text, keep.source = FALSE): <text>:1:9: unexpected symbol
#> 1: .dollar costs_min
#> ^
```

```r
epoxy("for a single child range from {.dollar costs_min$cost}")
```

```
#> for a single child range from $4,810
```

---
layout: false

## These are the same

.mt3[
### .Rmd

````markdown
```{epoxy}
for a single child range from {.dollar costs_min$cost}
```
````

for a single child range from $4,810
]

.mt4[
### .R

```r
epoxy("for a single child range from {.dollar costs_min$cost}")
```

```
#> for a single child range from $4,810
```
]

---
layout: true

## Inline transformations with epoxy

---

```r
epoxy(
  "range from {.dollar costs_min$cost} to {.dollar costs_max$cost}"
)
```

```
#> range from $4,810 to $15,416.96
```

---

```r
epoxy(
  "range from {.dollar costs_min$cost} to {.dollar costs_max$cost}",
  .transformer = epoxy_transform_inline()
)
```

```
#> range from $4,810 to $15,416.96
```

---

```r
epoxy(
  "range from {.dollar costs_min$cost} to {.dollar costs_max$cost}",
  .transformer = epoxy_transform_inline(
    .dollar = scales::label_dollar()
  )
)
```

```
#> range from $4,810 to $15,416.96
```

---

```r
epoxy(
  "range from {.dollar costs_min$cost} to {.dollar costs_max$cost}",
  .transformer = epoxy_transform_inline(
    .dollar = scales::label_dollar(accuracy = 10)
  )
)
```

```
#> range from $4,810 to $15,420
```

---

```r
epoxy_transform_set(
  .dollar = scales::label_dollar(accuracy = 10)
)

epoxy(
  "range from {.dollar costs_min$cost} to {.dollar costs_max$cost}"
)
```

```
#> range from $4,810 to $15,420
```

---
layout: true

## Inline transformations with epoxy

```r
ages <- childcare_summary |> count(age) |> mutate(across(age, paste))
```

<PRE class="fansi fansi-output"><CODE>#&gt; # A tibble: 4 × 2
#&gt; age n
#&gt; &lt;chr&gt; &lt;int&gt;
#&gt; 1 infant 8
#&gt; 2 toddler 8
#&gt; 3 preschool 8
#&gt; 4 school-age 8
</CODE></PRE>

---

```r
epoxy("children were in the age groups {ages$age}")
```

```
#> children were in the age groups infant
#> children were in the age groups toddler
#> children were in the age groups preschool
#> children were in the age groups school-age
```

---

```r
epoxy("children were in the age groups {.and ages$age}")
```

```
#> children were in the age groups infant, toddler, preschool, and school-age
```

---

```r
epoxy("children were in the age groups {.or ages$age}")
```

```
#> children were in the age groups infant, toddler, preschool, or school-age
```

---
layout: true

## Inline transformations with epoxy

---

```r
epoxy("age groups ranged from {.to ages$age}")
```

```
#> age groups ranged from infant
#> age groups ranged from toddler
#> age groups ranged from preschool
#> age groups ranged from school-age
```

---

```r
epoxy_transform_set(
  .to = function(x) {
    glue_collapse(x, sep = ", ", last = " to ")
  }
)

epoxy("age groups ranged from {.to ages$age}")
```

```
#> age groups ranged from infant, toddler, preschool to school-age
```

---

```r
epoxy("age groups ranged from {.to {.titlecase ages$age}}")
```

```
#> age groups ranged from Infant, Toddler, Preschool to School-Age
```

---

```r
epoxy("age groups ranged from {.to {.tc ages$age}}")
```

```
#> age groups ranged from Infant, Toddler, Preschool to School-Age
```

---

```r
epoxy(
  "These slides were rendered on {.date now()} at {.time now()}."
)
```

```
#> These slides were rendered on 2023-06-08 at 14:43:05.
```

.footnote.o-80[
`now()` is from `lubridate`, which is now loaded with `library(tidyverse)`!
]

---

```r
epoxy(
  "These slides were rendered on {.date now()} at {.time now()}.",
  .transformer = c("inline", "bold")
)
```

```
#> These slides were rendered on **2023-06-08** at **14:43:05**.
```

---
layout: true

## epoxy for .pink[LaTeX]

---

```r
epoxy("prices in {.comma n_counties} U.S. counties")
```

```
#> prices in 2,360 U.S. counties
```

```r
epoxy("prices in {{.comma n_counties}} U.S. counties")
```

```
#> prices in {.comma n_counties} U.S. counties
```

---

```r
epoxy("prices in \\textbf{{.comma n_counties}} U.S. counties")
```

```
#> prices in \textbf{.comma n_counties} U.S. counties
```

```r
epoxy("prices in \\textbf{{{.comma n_counties}}} U.S. counties")
```

```
#> prices in \textbf{2,360} U.S. counties
```

---

```r
epoxy_latex("prices in \\textbf{<.comma n_counties>} U.S. counties")
```

```
#> prices in \textbf{2,360} U.S. counties
```

This is equivalent to an `epoxy_latex` chunk...

````default
```{epoxy_latex}
prices in \\textbf{<.comma n_counties>} U.S. counties
```
````

...but the output of `epoxy_latex` chunks render in LaTeX outputs.

---
layout: true

## epoxy for .pink[HTML]

---

```r
epoxy_html("prices in {{.comma n_counties}} U.S. counties")
```

prices in 2360 U.S. counties

````html
prices in 2360 U.S. counties
````

---

```r
epoxy_html("prices in {{.comma n_counties}} U.S. counties")
```

prices in 2360 U.S. counties

````html
prices in 2360 U.S. counties
````

---

```r
epoxy_html("prices in {{strong n_counties}} U.S. counties")
```

prices in 2360 U.S. counties

````html
prices in 2360 U.S. counties
````

---

```r
epoxy_html("prices in {{strong {{.comma n_counties}}}} U.S. counties")
```

prices in 2,360 U.S. counties

````html
prices in 2,360 U.S. counties
````

---
layout: false
class: inverse center middle

# Templating small and large

---
layout: true

# Data-masked chunks

---

````markdown
```{epoxy}
{costs_min$cost} for {costs_min$age} {costs_min$type}-based care
in {costs_min$county_size} counties
```
````

4810 for school-age home-based care
in small counties

---

````markdown
*```{epoxy data=costs_min}
{.dollar costs_min$cost} for {costs_min$age} {costs_min$type}-based care
in {costs_min$county_size} counties
```
````

$4,810 for school-age home-based care
in small counties

---

````markdown
```{epoxy data=costs_min}
*{.dollar cost} for {age} {type}-based care
*in {county_size} counties
```
````

$4,810 for school-age home-based care
in small counties

---

````markdown
```{epoxy data=costs_min}
{.dollar cost} for {age} {type}-based care in {county_size} counties
```
````

$4,810 for school-age home-based care in small counties

````markdown
```{epoxy data=costs_max}
{.dollar cost} for {age} {type}-based care in {county_size} counties
```
````

$15,420 for infant center-based care in very large counties

---
layout: true

## Tiny templates with .pink[ref.label]

---

````markdown
```{epoxy cost-summary, data=costs_min}
{.dollar cost} for {age} {type}-based care in {county_size} counties
```
````

$4,810 for school-age home-based care in small counties

````markdown
```{epoxy ref.label = "cost-summary", data=costs_max}
```
````

$15,420 for infant center-based care in very large counties

---

````markdown
A study of childcare prices in
```{epoxy}
{.comma n_counties} U.S. counties across {n_states} states
```
shows that childcare prices for a single child range from
```
```{epoxy ref.label = "cost-summary", data=costs_min}
```
to
```{epoxy ref.label = "cost-summary", data=costs_max}
```
per year. This is equivalent to between
```{epoxy}
{.pct costs_min$mhi_pct} and {.pct costs_max$mhi_pct}
```
of median family income per child in paid care.
````

---

.w-two-thirds.mt4.lh-copy[
A study of childcare prices in
2,360 U.S. counties across 41 states
shows that childcare prices for a single child range from
$4,810 for school-age home-based care in small counties
to
$15,420 for infant center-based care in very large counties
per year. This is equivalent to between
10% and 24%
of median family income per child in paid care.
]

---
layout: false
class: inverse center middle

# &#x2728; Shiny Apps &#x2728;

---
layout: true

## epoxy in .pink[Shiny apps]

---

```r
radioButtons("county_size", "County Size", ...),
radioButtons("type", "Childcare Type", ...),
radioButtons("age", "Age Group", ...)
```

```r
p(
  "The median childcare cost for a single ",
  "AGE in TYPE-based care",
  "is COST in COUNTY_SIZE counties."
)
```

---

.o-50[
```r
radioButtons("county_size", "County Size", ...),
radioButtons("type", "Childcare Type", ...),
radioButtons("age", "Age Group", ...)
```
]

```r
ui_epoxy_html(
  .id = "summary",
  p(
    "The median childcare cost for a single ",
    "{{ age }} in {{ type }}-based care",
    "is {{ cost }} in {{ county_size }} counties."
  )
)
```

```r
output$summary <- render_epoxy(
 age = input$age,
 type = input$type,
 county_size = input$county_size,
 cost = ...
)
```

---

```r
output$summary <- render_epoxy(
 age = input$age,
 type = input$type,
 county_size = input$county_size,
 cost = cost()$cost
)
```

```r
cost <- reactive({
 childcare_summary |>
 filter(
 age == input$age,
 type == input$type,
 county_size == input$county_size
 ) |>
 mutate(cost = scales::dollar(cost, accuracy = 1))
})
```

---
layout: false
class: center middle

## Demo Time Field Trip

---
class: highlight-last-item

## But wait, .pink[there's more!]

* You can use `{{ }}` anywhere (almost) in any UI element!

.silver[Try .code[run_epoxy_example_app("ui_epoxy_html")]]

* You can use markdown instead with `ui_epoxy_markdown()`

.silver[Try .code[run_epoxy_example_app("ui_epoxy_markdown")]]

* Mustache templates with `ui_epoxy_mustache()` (and the `whisker` engine)

.silver[Try .code[run_epoxy_example_app("ui_epoxy_mustache")]]

---

## Thanks!

### Learn more about epoxy

-   [Get to know epoxy](https://pkg.garrickadenbuie.com/epoxy/) .silver[(.code[pkg.garrickadenbuie.com/epoxy])]

-   Would love to hear from you or help you use epoxy!
    Come to the [discussion board](https://github.com/gadenbuie/epoxy/discussions)

-   [slides.garricakdenbuie.com/seamless-epoxy](https://slides.garrickadenbuie.com/seamless-epoxy)

### About me

- Read: [garrickadenbuie.com](https://www.garrickadenbuie.com)

- Chat: [&commat;grrrck&commat;fosstodon.org](https://fosstodon.org/@grrrck)

- Code: [&commat;gadenbuie](https://github.com/gadenbuie)

.f3.mt4.blue[
&#x1F647;&#x200D;&#x2642;&#xFE0F; Thank you R/Medicine!
]