GraphTypes.qmd

---
title: "Graph types"
author: "Sean Raleigh <br> Westminster University"
execute: 
  echo: false
format:
    revealjs:
        theme: sky
        incremental: false
        embed-resources: true
---


# Data preparation

```{r}
library(palmerpenguins)
library(gt)
library(tidyverse)
library(ggbeeswarm)
library(viridis)
library(janitor)
library(tsibble)

theme_resize <- theme(title = element_text(size = rel(1.5)),
                      axis.title = element_text(size = rel(1.25)),
                      axis.text = element_text(size = rel(1)))
```


## Data preparation

- Make sure your data is "tidy":
    - Observations are rows.
    - Variables are columns.
    - Every cell is a single value.
    - First row (and only first row) consists of variable names.
    - No extra stuff outside the rectangle.

## Data preparation

Measurements for penguins in the Palmer Archipelago, Antarctica.

```{r}
penguins <- penguins %>%
    drop_na()
penguins %>%
    head(n = 20) %>%
    gt()
```


# Identifying variable types


## Identifying variable types

- Categorical (qualitative, nominal, factor)
    - Classifies data by category.
    - color, species, sex

- Numerical (quantitative, scale, interval/ratio)
    - Numerical measurements, usually with meaningful units.
    - height, GDP, score


## Identifying variable types

**CAREFUL!**

Numbers are not always numerical.

- *Do you own a car?*
    - 0 = "No", 1 = "Yes"
- *What is your zip code?*



# Single variable

- Single variables usually won't answer very interesting questions by themselves.
- Graphs of single variables are often valuable for exploring your data, but generally not suitable for inclusion in the final product.


## Single categorical variable

Penguin species.

```{r}
penguins %>%
    select(species) %>%
    head(n = 20) %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Single categorical variable

- Frequency table

```{r}
tabyl(penguins, species) %>%
    select(-percent) %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Single categorical variable

- Bar chart

```{r}
ggplot(penguins, aes(x = species)) +
    geom_bar() +
    labs(title = "Penguin species",
         x = "Species",
         y = "Count") +
    theme_resize
```


## Single categorical variable

- ~~Pie chart~~ **Danger! Danger!**

```{r}
penguins %>%
    count(species) %>%
    rename(Species = species) %>%
    ggplot(aes(x = "", y = n, fill = Species)) +
        geom_bar(stat = "identity", width = 1) +
        coord_polar("y") +
        theme_void() +
        scale_fill_viridis(discrete = TRUE)
```


## Single categorical variable

- Relative frequency table

```{r}
tabyl(penguins, species) %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Single categorical variable

<a href = "http://www.statschat.org.nz/wp-content/uploads/2015/06/devourThePie3.gif" target = "_blank">Click here to see how to improve a pie chart.</a>


## Single numerical variable

Penguin body mass in grams.

```{r}
penguins %>%
    select(body_mass_g) %>%
    head(n = 20) %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Single numerical variable

- Histogram

```{r}
ggplot(penguins, aes(x = body_mass_g)) +
    geom_histogram(color = "black", fill = "white",
                   binwidth =  250, boundary = 1000) +
    labs(title = "Penguin body mass (g)",
         x = "Body mass (g)",
         y = "Count") +
    theme_resize
```


## Single numerical variable

- Tabular summaries

"Five-number summary"

```{r}
penguins %>%
    reframe(Percentiles = c("0%", "25%", "50%", "75%", "100%"),
            Body_mass = quantile(body_mass_g)) %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Single numerical variable

- A bunch of other types I don't prefer:

> boxplot, stem-and-leaf plot, dotplot



# Multiple variables

There are at least six elements of a plot that can be assigned to variables:

- x-axis (horizontal axis)
- y-axis (vertical axis)
- facets
- color/fill
- size
- shape (e.g., dots vs crosses, solid vs dashed lines, etc.)


## Two categorical variables

Penguin species and island.

```{r}
penguins %>%
    select(species, island) %>%
    head(n = 20) %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Two categorical variables

- Contingency table (okay)

```{r}
tabyl(penguins, species, island) %>%
    adorn_totals() %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Two categorical variables

- Contingency table (better)

```{r}
tabyl(penguins, species, island) %>%
    adorn_totals() %>%
    adorn_percentages("col") %>%
    adorn_pct_formatting() %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Two categorical variables

- Side-by-side bar chart (okay)

```{r}
penguins %>%
    count(species, island) %>%
    bind_rows(tribble(~species, ~island, ~n, # Fix spacing
                      "Gentoo", "Torgersen", 0)) %>%
    ggplot(aes(x = island, y = n, fill = species)) +
        geom_bar(stat = "identity", position = "dodge") +
        labs(title = "Penguin species distribution by island",
         x = "Island",
         y = "Count",
         fill = "Species") +
    scale_fill_viridis(discrete = TRUE) +
    theme_resize
```


## Two categorical variables

- Side-by-side bar chart (better)

```{r}
penguins %>%
    count(species, island) %>%
    group_by(island) %>%
    mutate(prop = n / sum(n)) %>%
    bind_rows(tribble(~species, ~island, ~n, # Fix spacing
                      "Gentoo", "Torgersen", 0)) %>%
    ggplot(aes(x = island, y = prop, fill = species)) +
        geom_bar(stat = "identity", position = "dodge") +
        labs(title = "Penguin species distribution by island",
         x = "Island",
         y = "Count",
         fill = "Species") +
    scale_fill_viridis(discrete = TRUE) +
    theme_resize
```



## Two categorical variables

- Stacked bar chart (okay)

```{r}
ggplot(penguins, aes(x = island, fill = species)) +
    geom_bar() +
    labs(title = "Penguin species distribution by island",
         x = "Island",
         y = "Count",
         fill = "Species") +
    scale_fill_viridis(discrete = TRUE) +
    theme_resize
```

## Two categorical variables

- Stacked bar chart (okay)

```{r}
ggplot(penguins, aes(x = island, fill = species)) +
    geom_bar(position = "fill") +
    labs(title = "Penguin species distribution by island",
         x = "Island",
         y = "Count",
         fill = "Species") +
    scale_fill_viridis(discrete = TRUE) +
    theme_resize
```


## One categorical and one numerical variable

Penguin species and body mass (g).

```{r}
penguins %>%
    select(species, body_mass_g) %>%
    head(n = 20) %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## One categorical and one numerical variable

- Side-by-side boxplot (okay)

```{r}
ggplot(penguins, aes(x = species, y = body_mass_g)) +
    geom_boxplot() +
    labs(title = "Penguin body mass (g) by species",
         x = "Species",
         y = "Body mass (g)") +
    theme_resize
```


## One categorical and one numerical variable

- Stacked histogram (better)

```{r}
ggplot(penguins, aes(x = body_mass_g)) +
    geom_histogram(binwidth =  250, boundary = 1000,
                   color = "black",
                   fill = "white") +
    facet_grid(rows = vars(species)) +
    labs(title = "Penguin body mass (g) by species",
         y = "Count",
         x = "Body mass (g)") +
    theme_resize
```


## One categorical and one numerical variable

- ~~Dynamite plot~~ **Danger! Danger!**

```{r}
ggplot(penguins, aes(x = species, y = body_mass_g, fill = species)) +
    stat_summary(geom = "errorbar", fun.data = "mean_sdl",
                 width = .5, size = 1) +
    stat_summary(geom = "bar", fun.y = "mean") +
    labs(title = "Penguin body mass (g) by species",
         x = "Species",
         y = "Mean body mass (g)") +
    scale_fill_viridis(discrete = TRUE) +
    theme_resize
```

## One categorical and one numerical variable

- Beeswarm (better)

```{r}
ggplot(penguins, aes(y = body_mass_g, x = species)) +
    geom_quasirandom() +
    labs(title = "Penguin body mass (g) by species",
         y = "Count",
         x = "Body mass (g)") +
    theme_resize
```


## Two numerical variables

Penguin flipper length (mm) and body mass (g).

```{r}
penguins %>%
    select(flipper_length_mm, body_mass_g) %>%
    head(n = 20) %>%
    gt() %>%
    tab_options(table.font.size = "75%")
```


## Two numerical variables

- Scatterplot

```{r}
ggplot(penguins, aes(y = flipper_length_mm, x = body_mass_g)) +
    geom_point() +
    labs(title = "Penguin flipper length (mm) by\nbody mass (g)",
         y = "Flipper length (mm)",
         x = "Body mass (g)") +
    theme_resize
```


## Two numerical variables

- If one variable is ordered (like time) and there is only one observation of y for each x value, use a lineplot.

```{r}
ggplot(as_tsibble(co2), aes(y = value, x = index)) +
    geom_line() +
    labs(title = "Atmospheric CO2 in ppm (1959 to 1998)\nMauna Loa",
         x = "Year",
         y = "Atmospheric CO2 (ppm)\n") +
    theme_resize
```


## Three or more variables

```{r}
ggplot(penguins, aes(x = island, y = body_mass_g, fill = species)) +
    geom_boxplot() +
    labs(x = "Island",
         y =  "Body mass (g)",
         fill = "Species") +
    scale_fill_viridis(discrete = TRUE) +
    theme_resize
```


## Three or more variables

```{r}
ggplot(penguins, aes(x = body_mass_g, y = flipper_length_mm, color = sex)) +
    geom_point() +
    facet_grid(rows = vars(species),
               cols = vars(island)) +
    labs(y = "Flipper length (mm)",
         x = "Body mass (g)",
         color = "Sex") +
    scale_color_viridis(discrete = TRUE) +
    theme_dark() +
    theme_resize +
    theme(axis.text = element_text(size = rel(0.75)))
```