gt Tables

class: center, middle

## `gt` Tables

### Kelly McConville

.large[Math 241 | Week 7 | Spring 2021]

---

## Announcements/Reminders

* Lab 5 is posted.

---
  
## Goals for Today

* Finish up code smells and refactoring.

* A quick note on tidy evaluation.

* Talk about display tables.

---

## Non-Standard Evaluation

* On your lab this week, you are going to create functions for common `ggplot2` and `dplyr` operations.

* But that means we need to learn about one more idea

```r
library(pdxTrees)
pdxTrees_parks <- get_pdxTrees_parks()

# Minimal viable product working code
ggplot(data = pdxTrees_parks, mapping = aes(x = DBH)) +
  geom_histogram()
```

---

## Non-Standard Evaluation

* On your lab this week, you are going to create functions for common `ggplot2` and `dplyr` operations.

* But that means we need to learn about one more idea

```r
# Shorthand histogram function
histo <- function(data, x, ...){
  ggplot(data = data, mapping = aes(x = x)) +
    geom_histogram()
}

# Test it
histo(pdxTrees_parks, DBH)
```

```
## Error in FUN(X[[i]], ...): object 'DBH' not found
```

---

## Non-Standard Evaluation

* On your lab this week, you are going to create functions for common `ggplot2` and `dplyr` operations.

* But that means we need to learn about one more idea

```r
# Shorthand histogram function
histo <- function(data, x, ...){
  ggplot(data = data, mapping = aes(x = x)) +
    geom_histogram()
}

# Test it
histo(pdxTrees_parks, "DBH")
```

```
## Error: StatBin requires a continuous x variable: the x variable is discrete.Perhaps you want stat="count"?
```

---

## Non-Standard Evaluation

**Solution 1**: Supply the column name as a character vector and then use `.data[[ x ]]` to locate it.

```r
histo <- function(data, x, ...){
  ggplot(data = data, mapping = aes(x = .data[[x]])) +
    geom_histogram(...)
}

# Test it
histo(pdxTrees_parks, "DBH")
```

---

**Solution 1**: Supply the column name as a character vector and then use `.data[[ x ]]` to locate it.

```r
histo <- function(data, x, ...){
  stopifnot(is.numeric(data[[x]]))
  ggplot(data = data, mapping = aes(x = .data[[x]])) +
    geom_histogram(...)
}

# Test it
histo(pdxTrees_parks, "Condition")
```

```
## Error in histo(pdxTrees_parks, "Condition"): is.numeric(data[[x]]) is not TRUE
```

---

## Non-Standard Evaluation

**Solution 2**: If you pass the column name directly, use `{{ x }}` to pass it to a tidy evaluation enabled function.

```r
histo <- function(data, x, ...){
  ggplot(data = data, mapping = aes(x = {{ x }})) +
    geom_histogram()
}

# Test it
histo(pdxTrees_parks, DBH)
```

---

## Non-Standard Evaluation

**Solution 2**: If you pass the column name directly, use `{{ x }}` to pass it to a tidy evaluation enabled function.

```r
histo <- function(data, x, ...){
  stopifnot(is.numeric(rlang::eval_tidy(enquo(x), data)))
  ggplot(data = data, mapping = aes(x = {{ x }})) +
    geom_histogram()
}

# Test it
histo(pdxTrees_parks, Condition)
```

```
## Error in histo(pdxTrees_parks, Condition): is.numeric(rlang::eval_tidy(enquo(x), data)) is not TRUE
```

---

## [Display Format](https://www.bts.gov/content/age-and-availability-amtrak-locomotive-and-car-fleets)

Unfortunately, many datasets on the internet are often in **Display Format**, not **Analysis Format**.

---

## [Display Format](https://www.bts.gov/content/age-and-availability-amtrak-locomotive-and-car-fleets)

This means we spend a LOT of energy wrangling the data so that we can use it in our analyses.

```r
library(tidyverse)
library(readxl)

Amtrak <- read_excel("table_01_33_102020.xlsx", range = c("A2:AI8"),
                     na = c("U", NA)) %>%
  rename(Vars = ...1) %>%
  slice(-1, -4) %>%
  mutate(TrainType = c(rep("Locomotives", 2),
                       rep("Passenger and Other", 2))) %>%
  relocate(TrainType) %>%
  mutate(Vars = case_when(
    Vars == "Percent available for servicea" ~ "PercentAvailable",
    Vars == "Average age (years)b" ~ "AverageAge"
  )) %>%
  pivot_longer(cols = !c(TrainType, Vars), names_to = "Year", values_to = "Value") %>%
  pivot_wider(names_from = Vars, values_from = Value) %>%
  mutate(Year = as.numeric(Year))
```

---

## Display Format

This means we spend a LOT of energy wrangling the data so that we can use it in our analyses.

```r
Amtrak 
```

```
## # A tibble: 68 x 4
##    TrainType    Year PercentAvailable AverageAge
##    <chr>       <dbl>            <dbl>      <dbl>
##  1 Locomotives  1972               NA       22.3
##  2 Locomotives  1975               87       14.4
##  3 Locomotives  1980               83        7.4
##  4 Locomotives  1985               93        7  
##  5 Locomotives  1990               84       12  
##  6 Locomotives  1991               86       13  
##  7 Locomotives  1992               83       13  
##  8 Locomotives  1993               84       13.2
##  9 Locomotives  1994               85       13.4
## 10 Locomotives  1995               88       13.9
## # … with 58 more rows
```

---

## Display Format

This means we spend a LOT of energy wrangling the data so that we can use it in our analyses.

```r
ggplot(data = Amtrak, mapping = aes(x = Year, y = AverageAge,
                                    color = TrainType)) +
  geom_line()
```

---

## Display Format

But once you have finished analyzing the data, you may want to now convert a table back to **display format**.

---

## Display Format

And, just printing the table doesn't really cut it!

```r
Amtrak %>% 
  filter(Year <= 2015, Year > 2010) 
```

```
## # A tibble: 10 x 4
##    TrainType            Year PercentAvailable AverageAge
##    <chr>               <dbl>            <dbl>      <dbl>
##  1 Locomotives          2011             84.2       20  
##  2 Locomotives          2012             83.7       21  
##  3 Locomotives          2013             80.3       21.9
##  4 Locomotives          2014             82.5       21.5
##  5 Locomotives          2015             82         21.1
##  6 Passenger and Other  2011             87.7       26.5
##  7 Passenger and Other  2012             88.7       27.7
##  8 Passenger and Other  2013             89.1       28.6
##  9 Passenger and Other  2014             89.1       29.6
## 10 Passenger and Other  2015             88.8       30.7
```

---

## But Aren't Graphs > Tables for Displaying Data?

Graphs are useful when:

* You want to display trends, relationships, and/or overall shape.
* You want to pull in the potential reader/viewer.

Tables are useful when:

* You want to provide structured, summarized and aggregated information.
* You want to show the exact values.
* The viewer wants to be able to look up and compare individual values.

Like with graphs, you should

* Format your tables to suit their function.
* Make important comparisons easy.

---

## `gt` (Grammar of Tables) Package

.left-column[

]

.right-column[

]

---

class: inverse, middle, center

### Handout: gt.Rmd