overlapping_stacked_density_plots.qmd

# Overlapping density plot {#sec-overlapping-stacked-density-plots}

```{r}
#| label: setup
#| message: false
#| warning: false
#| include: false
library(tidyverse)
library(lubridate)
library(scales)
library(knitr)
library(kableExtra)
library(colorblindr)
library(downlit)
# fonts ----
library(extrafont)
library(sysfonts)
source("_common.R")
# use font
showtext::showtext_auto()
# set theme
ggplot2::theme_set(theme_ggp2g(
    base_size = 15))
```

```{r}
#| label: status
#| results: "asis"
#| echo: false
# status ----
# polish, dev, draft, complete
status("complete")
```



:::: {.callout-note collapse="false" icon=false}

<br>

```{r}
#| label: full_code_display
#| eval: true
#| echo: false
#| warning: false
#| message: false
#| out-height: '60%'
#| out-width: '60%'
#| fig-align: right
library(palmerpenguins) 
library(ggplot2)
library(patchwork)
peng_density <- filter(penguins, !is.na(sex))
labs_ovrlp_density <- labs(
  subtitle = "position = 'identity'",
  x = "Flipper length (mm)",
  fill = "Sex")
ggp2_ovrlp_density <- ggplot(data = peng_density, 
       aes(x = flipper_length_mm, 
           fill = sex)) +
      geom_density(alpha = 1/3) 

ovrlp_density <- ggp2_ovrlp_density + 
  labs_ovrlp_density + 
    ggplot2::theme_minimal(base_size = 11)

labs_stack_density <- labs(
  subtitle = "position = 'stack'",
  x = "Flipper length (mm)", 
  fill = "Sex")
ggp2_stack_density <- ggplot(data = peng_density, 
       aes(x = flipper_length_mm, 
           after_stat(count),
           fill = sex)) +
      geom_density(position = "stack", 
                   alpha = 1/3) 
stack_density <- ggp2_stack_density + 
  labs_stack_density + 
    ggplot2::theme_minimal(base_size = 11)
ovrlp_density + stack_density + 
    patchwork::plot_layout(ncol = 1)
```

::: {style="font-size: 1.10em; color: #02577A;"}
**This graph requires:**
:::

::: {style="font-size: 1.05em; color: #282b2d;"}
**`r emo::ji("check")` a categorical variable**      
:::

::: {style="font-size: 1.05em; color: #282b2d;"}
**`r emo::ji("check")` a numeric (continuous) variable**
:::

::::

## Description 

Density plots are [*smoothed version(s) of  histogram(s)*](https://ggplot2.tidyverse.org/reference/geom_density.html). They can are great for comparing the distributions of a continuous variable across the levels of a categorical variable. 

`geom_density()` creates a kernel density estimate. The default `position` argument is `"identity"`, which takes the data as is. However, we can change `position` to `"stack"` to display overlapping distributions.

## Set up

::: {style="font-size: 1.15em; color: #1e83c8;"}
**PACKAGES:**
:::

::: {style="font-size: 0.95rem;"}
Install packages.
:::


::: {style="font-size: 0.85em;"}
```{r}
#| label: pkg_code_ovrlp_density
#| code-fold: show
#| eval: true
#| echo: true
#| warning: false
#| message: false
#| results: hide
install.packages("palmerpenguins")
library(palmerpenguins)
library(ggplot2)
```
:::



::: {style="font-size: 1.15em; color: #1e83c8;"}
**DATA:**
:::

::: {.column-margin}
![Artwork by allison horst](www/lter_penguins.png){fig-align="right" width="100%" height="100%"}
:::

::: {style="font-size: 0.95rem;"}
Remove missing `sex` from the `penguins` data
:::

::: {style="font-size: 0.85em;"}
```{r}
#| label: data_code_ovrlp_density
#| code-fold: show
#| eval: true
#| echo: true
peng_density <- dplyr::filter(penguins, !is.na(sex))
dplyr::glimpse(peng_density)
```
:::

## Grammar

::: {style="font-size: 1.15em; color: #1e83c8;"}
**CODE:**
:::



  -   Create labels with `labs()`

  -   Initialize the graph with `ggplot()` and provide `data`

  -   Map the `flipper_length_mm` to the `x` and `sex` to `fill` 

  -   Add the `geom_density()`  

  -   Set the `alpha` to `1/3` (to handle the overlapping areas)

::: {style="font-size: 0.85em;"}
```{r}
#| label: code_graph_ovrlp_density
#| code-fold: show
#| eval: false
#| echo: true 
#| warning: false
#| message: false
#| out-height: '100%'
#| out-width: '100%'
labs_ovrlp_density <- labs(
  title = "Adult foraging penguins",
  x = "Flipper length (millimeters)",
  fill = "Sex")
ggp2_ovrlp_density <- ggplot(data = peng_density, 
       aes(x = flipper_length_mm, 
           fill = sex)) +
      geom_density(alpha = 1/3) 
ggp2_ovrlp_density + 
  labs_ovrlp_density
```
:::



::: {style="font-size: 1.15em; color: #1e83c8;"}
**GRAPH:**
:::

::: {.column-margin}
*A downside of density plots is the lack of interpretability of the `y` axis*
:::


Make density area slightly transparent to handle over-plotting.

```{r}
#| label: create_graph_ovrlp_density
#| eval: true
#| echo: false
#| warning: false
#| message: false
#| out-height: '100%'
#| out-width: '100%'
#| layout-nrow: 1
#| column: page-inset-right
labs_ovrlp_density <- labs(
  title = "Adult foraging penguins",
  x = "Flipper length (millimeters)",
  fill = "Sex")
ggp2_ovrlp_density <- ggplot(data = peng_density, 
       aes(x = flipper_length_mm, 
           fill = sex)) +
      geom_density(alpha = 1/3) 
ggp2_ovrlp_density + 
  labs_ovrlp_density
```

## More info

`ggplot2` has multiple options for overlapping density plots, so which one to use will depend on how you'd like to display the relative distributions in your data. We'll cover three options below:

### `"stack"`

If we change the `position` to `"stack"` we can see the smoothed estimates are 'stacked' on top each other (and the `y` axis shifts slightly).

::: {style="font-size: 0.85em;"}

```{r}
#| label: code_graph_stack_density
#| eval: true
#| echo: true 
#| warning: false
#| out-height: '100%'
#| out-width: '100%'
#| layout-nrow: 1
#| column: page-inset-right
labs_stack_density <- labs(
    title = "Adult foraging penguins",
    x = "Flipper length (millimeters)",
    fill = "Sex")
ggp2_stack_density <- ggplot(data = peng_density,
    mapping = aes(x = flipper_length_mm,
               fill = sex)) +
    geom_density(position = "stack",
                 alpha = 1 / 3)
ggp2_stack_density +
    labs_stack_density
```
:::

::: {style="font-size: 0.95rem;"}
Setting `position` to `'stack'` [loses marginal densities](https://ggplot2.tidyverse.org/reference/geom_density.html#ref-examples)
:::

### `after_stat(count)`


If we include `after_stat(count)` as one of our mapped aesthetics, the mapping is postponed until after statistical transformation, and uses the [*`density * n` instead of the default density*](https://ggplot2.tidyverse.org/reference/geom_density.html#computed-variables).

::: {style="font-size: 0.85em;"}
```{r}
#| label: code_graph_after_stat_density
#| eval: true
#| echo: true 
#| warning: false
#| message: false
#| out-height: '100%'
#| out-width: '100%'
#| layout-nrow: 1
#| column: page-inset-right
labs_after_stat_density <- labs(
  title = "Adult foraging penguins",
  x = "Flipper length (millimeters)", 
  fill = "Sex")
ggp2_after_stat_density <- ggplot(data = peng_density, 
       aes(x = flipper_length_mm, 
           after_stat(count),
           fill = sex)) +
      geom_density(position = "stack", 
                   alpha = 1/3) 
ggp2_after_stat_density + 
  labs_after_stat_density
```
:::


Adding [`after_stat(count)`](https://ggplot2.tidyverse.org/reference/geom_density.html#computed-variables) *'preserves marginal densities.'*, which result in more a interpretable `y` axis (depending on the audience)

### `"fill"`


Using `after_stat(count)` with `position = "fill"` creates in a conditional density estimate. 

::: {style="font-size: 0.85em;"}
```{r}
#| label: code_graph_fill_density
#| eval: true
#| echo: true 
#| warning: false
#| message: false
#| out-height: '100%'
#| out-width: '100%'
#| layout-nrow: 1
#| column: page-inset-right
labs_fill_density <- labs(
  title = "Adult foraging penguins",
  x = "Flipper length (millimeters)", 
  fill = "Sex")
ggp2_fill_density <- ggplot(data = peng_density, 
       aes(x = flipper_length_mm, 
           after_stat(count),
           fill = sex)) +
      geom_density(position = "fill", 
                   alpha = 1/3) 
ggp2_fill_density + 
  labs_fill_density
```
:::


This results in a `y` axis ranging from 0-1, and the area filled with the relative proportional values.