faceting.qmd

# Faceting

In this chapter, we will introduce faceting, a technique that produces multiple plots in which the data is subsetted by a categorical feature.

## Faceting on one variable: `facet_wrap()`

```{r}
#| echo: false
library(tidyverse)
```

```{r}
#| fig-width: 5
#| fig-height: 3

ggplot(iris, aes(Sepal.Length, Sepal.Width)) +  
  geom_point(color = "cornflowerblue") +  
  facet_wrap(~Species)
```

To control the layout dimensions, you can choose the number of rows or columns (`nrow =` or `ncol =`).

## Reordering facets

Facets are ordered according to the factor level order of the faceted variable. See [the chapter on factors](factors.hmtl) for strategies on reordering factor levels. 

```{r}
library(MASS)
painters |> 
  rownames_to_column("Name") |> 
  filter(School == "A") |> 
  pivot_longer(Composition:Expression, names_to = "Skill",
               values_to = "Score") |> 
  ggplot(aes(x = Score, y = fct_reorder(Name, Score))) +
  geom_point(color= "#786FB8") +
  facet_wrap(~Skill, ncol = 1)
  
  
  
  

```


```{r}
birds_plot <- birds |> 
  group_by(phase_of_flt) |> 
  summarize(count = n()) |> 
  slice_max(order_by = count, n = 4)
  
ggplot(birds, aes(x = speed, y = after_stat(density))) + 
  geom_histogram() +
  facet_wrap(~phase_of_flt, nrow = 1)
```


## Faceting on two variables: `facet_grid()`

With two variables, the rows represent the levels of one variable and the columns the other. These can be specified with the formula notation:
`facet_wrap(`*row variable*` ~ `*column variable*`)`.

```{r}
library(scales)
library(openintro)
ggplot(cle_sac, aes(x = age, y = personal_income)) +
  geom_point(size = 1, color = "cornflowerblue") +
  facet_grid(sex ~ city) +
  scale_y_continuous(labels = unit_format(unit = "K", scale = .001)) +
  labs(x = "age (in years)", y = "personal income", 
       caption = "Data: openintro::cle_sac") +
  theme_bw(16)
```

## Faceting with categorical variables


```{r}
#| fig-width: 10
#| fig-height: 6
#| out-width: 90```
library(pgmm) 
data(wine)  
tidywine <- wine |> 
  pivot_longer(cols = -Type, names_to = "variable",
               values_to = "value")  
tidywine |>
  ggplot(aes(value)) +
  geom_histogram(color = "blue", fill = "cornflowerblue",
                 bins = 20) +
  facet_wrap(~variable) +
  labs(title = "Fixed scales (default)") +
  theme_grey(14)
```

Axis scales can be made independent, by setting scales to `free`, `free_x`, or `free_y`. 

In this case, `scales = "free_x"` is a better option because the distribution of each numerical variable is more obvious.

```{r}
#| fig-width: 10
#| fig-height: 6
#| out-width: 90```
tidywine |>
  ggplot(aes(value)) +
  geom_histogram(color = "blue", fill = "cornflowerblue",
                 bins = 20) +
  facet_wrap(~variable, scales = "free_x") +
  labs(title = "Free x scale") +
  theme_grey(14)
```

##  Faceting on two variables

`facet_grid` can be used to split data-sets on two variables and plot them on the horizontal and/or vertical direction.

```{r,fig.width=10, fig.height=6}
wine |>   
  mutate(Type = paste("Type", Type)) |>   
  select(1:6) |>   
  pivot_longer(cols = -Type, names_to = "variable", values_to = "value") |>  
  ggplot(aes(value)) +  
  geom_histogram(color = mycol, fill = "lightblue") +  
  facet_grid(Type ~ variable, scales = "free_x") +  
  theme_grey(14)
```