featurette

class: left
background-image: url(https://images.unsplash.com/photo-1533930086187-0fc58e5a92e2?q=80&w=1587&auto=format&fit=crop&ixlib=rb-4.0.3&ixid=M3wxMjA3fDB8MHxwaG90by1wYWdlfHx8fGVufDB8fHx8fA%3D%3D)
background-size: cover
# Prediction with [{ggsmoothfit}](https://github.com/EvaMaeRey/ggsmoothfit)

Gina Reynolds | 2023-10-25 |Image credit: Elena Koycheva, Upsplash

### stat_smooth()'s xseq argument specifies positions at which to make predictions
--

### {ggsmoothfit} piggybacks on StatSmooth, to easily draw fitted values, residuals, and more.

---

```r
library(ggchalkboard)
library(tidyverse)

theme_minimal <- function(base_size = 11, 
                          base_family = "", 
                          base_line_size = base_size/22, 
                          base_rect_size = base_size/22,
                          board_color = "darkseagreen4", chalk_color = "lightyellow") {
    theme_bw(
      base_size = base_size, 
      base_family = base_family, 
      base_line_size = base_line_size, 
      base_rect_size = base_rect_size
    ) %+replace% 
    theme(
      rect = ggplot2::element_rect(fill = board_color),
      text = ggplot2::element_text(color = chalk_color, face = "italic", size = 15),
      panel.background = ggplot2::element_rect(fill = board_color),
      panel.border = element_blank(),
      legend.key = ggplot2::element_blank(),
      legend.title = ggplot2::element_blank(),
      axis.text = ggplot2::element_text(color = chalk_color),
      axis.ticks = ggplot2::element_line(color = chalk_color),
      panel.grid = ggplot2::element_blank(), 
      complete = TRUE
    )
}

theme_set(theme_minimal())

color = "lightyellow"; fill = color

ggplot2::update_geom_defaults("point", list(colour = color, 
        size = 2.5, alpha = 0.75))
    ggplot2::update_geom_defaults("segment", list(colour = color, 
        linewidth = 1.25, alpha = 0.75))
    ggplot2::update_geom_defaults("rug", list(colour = color, 
        size = 1, alpha = 0.75))
    ggplot2::update_geom_defaults("rect", list(colour = color, 
        size = 1, alpha = 0.75))
    ggplot2::update_geom_defaults("label", list(fill = fill, 
        color = "grey35", size = 5))
```

---

count: false

.panel1-feature-auto[

```r
*ggplot(mtcars) +  aes(wt, mpg)
```
]
 
.panel2-feature-auto[
![](ggsmoothfit_files/figure-html/feature_auto_01_output-1.png)
]

---
count: false

.panel1-feature-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
* geom_point()
```
]
 
.panel2-feature-auto[
![](ggsmoothfit_files/figure-html/feature_auto_02_output-1.png)
]

---
count: false

.panel1-feature-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
  geom_point() +
* geom_smooth(se = F, method = lm)
```
]
 
.panel2-feature-auto[
![](ggsmoothfit_files/figure-html/feature_auto_03_output-1.png)
]

---
count: false

.panel1-feature-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
  geom_point() +
  geom_smooth(se = F, method = lm) +
* ggsmoothfit:::geom_fit(method = lm)
```
]
 
.panel2-feature-auto[
![](ggsmoothfit_files/figure-html/feature_auto_04_output-1.png)
]

---
count: false

.panel1-feature-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
  geom_point() +
  geom_smooth(se = F, method = lm) +
  ggsmoothfit:::geom_fit(method = lm) +
* ggsmoothfit:::geom_residuals(method = lm)
```
]
 
.panel2-feature-auto[
![](ggsmoothfit_files/figure-html/feature_auto_05_output-1.png)
]

---

count: false

.panel1-feature2-auto[

```r
*ggplot(mtcars) +  aes(wt, mpg)
```
]
 
.panel2-feature2-auto[
![](ggsmoothfit_files/figure-html/feature2_auto_01_output-1.png)
]

---
count: false

.panel1-feature2-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
* geom_point() + geom_smooth(se = F, method = lm)
```
]
 
.panel2-feature2-auto[
![](ggsmoothfit_files/figure-html/feature2_auto_02_output-1.png)
]

---
count: false

.panel1-feature2-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
  geom_point() + geom_smooth(se = F, method = lm) +
* ggsmoothfit:::geom_smooth_predict(
*   xseq = 0, method = lm, size = 5)
```
]
 
.panel2-feature2-auto[
![](ggsmoothfit_files/figure-html/feature2_auto_03_output-1.png)
]

---
count: false

.panel1-feature2-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
  geom_point() + geom_smooth(se = F, method = lm) +
  ggsmoothfit:::geom_smooth_predict(
    xseq = 0, method = lm, size = 5) +
* ggsmoothfit:::geom_smooth_predict(
*   xseq = 2, method = lm, size = 5)
```
]
 
.panel2-feature2-auto[
![](ggsmoothfit_files/figure-html/feature2_auto_04_output-1.png)
]

---
count: false

.panel1-feature2-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
  geom_point() + geom_smooth(se = F, method = lm) +
  ggsmoothfit:::geom_smooth_predict(
    xseq = 0, method = lm, size = 5) +
  ggsmoothfit:::geom_smooth_predict(
    xseq = 2, method = lm, size = 5) +
* ggsmoothfit:::geom_smooth_predict(
*   xseq = 3, method = lm, size = 5)
```
]
 
.panel2-feature2-auto[
![](ggsmoothfit_files/figure-html/feature2_auto_05_output-1.png)
]

---
count: false

.panel1-feature2-auto[

```r
ggplot(mtcars) +  aes(wt, mpg) +
  geom_point() + geom_smooth(se = F, method = lm) +
  ggsmoothfit:::geom_smooth_predict(
    xseq = 0, method = lm, size = 5) +
  ggsmoothfit:::geom_smooth_predict(
    xseq = 2, method = lm, size = 5) +
  ggsmoothfit:::geom_smooth_predict(
    xseq = 3, method = lm, size = 5) +
* ggsmoothfit:::geom_smooth_step(
*   xseq = 2:3, method = lm)
```
]
 
.panel2-feature2-auto[
![](ggsmoothfit_files/figure-html/feature2_auto_06_output-1.png)
]

---

### Contribute

- https://github.com/EvaMaeRey/ggsmoothfit

---

### Check out stat_smooth() which does the computation in ggsmoothfit

- https://ggplot2.tidyverse.org/reference/geom_smooth.html
- examples with xseq, https://evamaerey.github.io/featurette/2023-10-30-stat-smooth-mtcars/stat-smooth-mtcars.html#10
- xseq documentation discussion https://github.com/tidyverse/ggplot2/issues/5246

---

### Check out flipbookr, used to build this featurette

- https://github.com/EvaMaeRey/flipbookr
- discussion: https://github.com/EvaMaeRey/flipbookr/blob/master/docs/draft_jasa_submission.pdf

---

### Check out more featurettes

- https://EvaMaeRey.github.io/featurette