Chapter15.qmd

---
title: "Chapter 15"
subtitle: "Coordinate Systems"
author: "Aditya Dahiya"
date: "2024-03-04"
format: 
  html: 
    code-fold: true
    code-copy: hover
    code-link: true
execute: 
  echo: true
  warning: false
  error: false
filters:
  - social-share
share:
  permalink: "https://aditya-dahiya.github.io/ggplot2book3e/Chapter15.html"
  description: "Solutions Manual (and Beyond) for ggplot2: Elegant Graphics for Data Analysis (3e)"
  twitter: true
  facebook: true
  linkedin: true
  email: true
  mastodon: true
editor_options: 
  chunk_output_type: console
bibliography: references.bib
---

::: {.callout-note appearance="minimal"}
This chapter has no exercises, so I use a dummy dataset to generate some examples of the functions and ggplot2 customization of coordinates discussed in this chapter.
:::

```{r}
#| label: setup

library(tidyverse)
library(scales)
library(gt)
library(gtExtras)
```

To begin with, lets create a dummy data set shown in @fig-tb

```{r}
#| label: fig-tb
#| fig-cap: "A dummy dataset to use in the examples for this Chapters' Solutions / Examples"

set.seed(123)
tb <- tibble(
  v_integer = 1:20,
  v_squared = v_integer ^ 2,
  v_regress = round(
    (v_integer ^ 2) + 
      rnorm(
        20, mean = 0, sd = 1), 
    2),
  v_random = sample(1:100, 20, replace = FALSE),
  v_discrete = sample(LETTERS[1:4], 20, replace = TRUE)
)
tb |> 
  gt() |> 
  cols_align("center") |> 
  gt_theme_538()
```

## 15.1 Linear coordinate systems

Let's use the example plot in @fig-sec15 to demonstrate the various changes / customization possible with ggplot2 with coordinates.

```{r}
#| label: fig-sec15
#| fig-cap: "An example plot which we will later use in various examples of tweaking the Linear Coordinate Systems"

g1 <- tb |>
  ggplot(
    aes(
      x = v_integer,
      y = v_regress
    )
  ) +
  geom_smooth(
    se = TRUE,
    colour = "grey",
    fill = "lightgrey",
    alpha = 0.5
  ) +
  geom_point(
    aes(colour = v_discrete),
    size = 3
  ) +
  scale_colour_brewer(palette = "Dark2") +
  theme_minimal() +
  theme(axis.line = element_line(arrow = arrow()))
g1
```

### 15.1.1 Zooming into a plot with `coord_cartesian()`

The @fig-zoom shows an example of zooming into a plot with `coord_cartesian()` and its difference from the other option using `scale_*_continuous(limits = *)` .

```{r}
#| label: fig-zoom
#| fig-cap: "Zooming in on a specified portion of Y-Axis: 100 to 200"
#| fig-subcap: 
#|   - "Base Plot"
#|   - "Limits argument in scale_y_continuous()"
#|   - "ylim argument in coord_cartesian()"

g1 +
  labs(title = "The Base Plot shows entire data")

g1 + scale_y_continuous(limits = c(100, 200)) +
  labs(title = "Using limits in scale_y_continuous() eliminates\nall data out of limits")

g1 + coord_cartesian(ylim = c(100, 200)) +
  labs(title = "Using ylim in coord_cartesian() simply zooms-in,\nand preserves out-of-limits data")


```

### 15.1.2 Flipping the axes with `coord_flip()`

The @fig-flip shows examples of flipping the coordinates, and reasoning as to why it may be better than changing X-axis and Y-axis variables manually, especially if we are using `geom_smooth()` which uses `y ~ x` formula by default.

```{r}
#| label: fig-flip
#| fig-cap: "Demonstrating the use of coord_flip()"
#| fig-subcap: 
#|   - "The base plot"
#|   - "Smoother line still follows v_regress ~ v_integer formula"
#|   - "Smoother line now follows v_integer ~ v_regress formula"

g1 + labs(title = "The base plot",
                subtitle = "Dependent variable is v_regress, and independent var is v_integer")

g1 + 
  coord_flip() +
  labs(title = "coord_flip()",
       subtitle = "Smoother line still follows v_regress ~ v_integer formula")

tb |>
  ggplot(
    aes(
      y = v_integer,
      x = v_regress
    )
  ) +
  geom_smooth(
    se = TRUE,
    colour = "grey",
    fill = "lightgrey",
    alpha = 0.5
  ) +
  geom_point(
    aes(colour = v_discrete),
    size = 3
  ) +
  scale_color_brewer(palette = "Dark2") +
  theme_minimal() +
  theme(axis.line = element_line(arrow = arrow())) + 
  labs(title = "Exchanging x and y variable",
       subtitle = "Smoother line now follows v_integer ~ v_regress formula")
```

### 15.1.3 Equal scales with `coord_fixed()`

The @fig-coordfixed shows the same three graphs above, but in a much better sense now that coordinates are fixed to be equal.

```{r}
#| label: fig-coordfixed
#| fig-cap: "Demonstrating the use of coord_fixed()"
#| fig-subcap: 
#|   - "The base plot"
#|   - "The base plot with fixed coordinates"
#| layout-ncol: 2

g1 + labs(title = "The base plot")

g1 + coord_fixed() + labs(title = "The base plot with coord_fixed()")

```

## 15.2 Non-linear coordinate systems

The @fig-coord1 demonstrates five types of coordinate systems - first three linear, and the last two non-linear.

```{r}
#| label: fig-coord1
#| fig-cap: "A demonstration of Polar Coordinates, Flipped Coordinates, Transformed Coordinates and Fixed Coordinates"
#| fig-subcap: 
#|   - "coord_cartesian()"
#|   - "coord_flip()"
#|   - "coord_fixed()"
#|   - "coord_polar()"
#|   - "coord_trans()"

tb <- tibble(
  x_random = runif(100, min = 0, max = 100),
  y_random = (1.5 * x_random) + (rnorm(100, 0, 25))
)

g1 <- ggplot(tb, aes(x_random, y_random)) +
  geom_point() +
  geom_smooth(se = FALSE) +
  labs(title = "Cartesian Coordinates") +
  theme_minimal()

g1 

g1 + coord_flip() + 
  labs(title = "Flipped Coordinates",
       subtitle = "coord_flipped()")

g1 + coord_fixed() +
  labs(title = "Fixed Coordinates",
       subtitle = "coord_fixed()")

g1 + coord_polar() +
  labs(title = "Polar Coordinates",
       subtitle = "coord_polar() transforms x-axis into theta (angle) and y-axis into radius")

g1 + coord_trans(x = "log10") +
  labs(title = "Transformed Coordinates",
       subtitle = "coord_trans() with x-axis on log10 scale")
```

### 15.2.1 Transformations with `coord_trans()`

The impact of `coord_trans()` can be demonstrated in @fig-c-trans shown below.

```{r}
#| label: fig-c-trans
#| fig-cap: "Use of coord_trans() to change the graph into a non-linear one"

g1 +
  coord_trans(x = "log2") +
  scale_x_continuous(breaks = 2^(1:8)) +
  theme(panel.grid.minor = element_blank()) +
  labs(
    title = "Transformation to Log 2 scale",
    subtitle = "Note that x-axis has equidistant powers of 2"
  )
```

### 15.2.2 Polar coordinates with `coord_polar()`

The @fig-polar demonstrates various uses of `coord_polar()`

```{r}
#| label: fig-polar
#| fig-cap: "Using polar coordinates"
#| fig-subcap: 
#|   - "A Stacked bar chart"
#|   - "Chart with coord_polar()"
#|   - "Bulls-Eye Chart"
#|   - "Pie-Chart"

tb2 <- tibble(
  Type = sample(LETTERS[1:4], size = 10, replace = T)
)

g2 <- tb2 |> 
  ggplot(aes(x = "1", fill = Type)) +
  geom_bar() +
  labs(title = "A Stacked Bar-Chart") +
  theme_void() +
  theme(legend.position = "bottom")

g2


g2 + 
  coord_polar() +
  labs(title = "coord_polar()",
       subtitle = "Without use of expand = expansion(0)")

g2 + 
  coord_polar() +
  scale_x_discrete(expand = expansion(0)) +
  labs(title = "Bulls-Eye Chart",
       subtitle = "Using expand = expansion(0) in scale_x_discrete()")

g2 + 
  coord_polar(theta = "y") +
  labs(title = "Pie Chart",
       subtitle = "Using theta = 'y' in coord_polar()")
```

### 15.2.3 Map projections with `coord_map()`

Here, in @fig-map, we show the map of Canada with different projections.

```{r}
#| label: fig-map
#| fig-cap: "Map of Canada with different coordinates"
#| fig-subcap: 
#|   - "Base map in coord_cartesian()"
#|   - "coord_quickmap()"
#|   - "coord_map()"
#|   - "coord_map(projection = `ortho`)"

g3 <- map_data(
  database = "world",
  regions = "canada"
) |> 
  ggplot(
    aes(
      x = long,
      y = lat,
      group = group
    )
  ) +
  geom_polygon(
    fill = "white",
    colour = "black"
  ) +
  theme(legend.position = "none") +
  labs(title = "Canada's Map: coord_cartesian()")

g3 + 
  coord_quickmap() +
  labs(title = "Canada's Map: coord_quickmap()")

g3 + 
  coord_map() +
  labs(title = "Canada's Map: coord_map()")

g3 + 
  coord_map("ortho") +
  labs(title = "Canada's Map: coord_map(projection = `ortho`)")

```