datavisualization.qmd

# Data visualization

\

## Creating graphs with `ggplot()`

`gg`stands for [grammar of graphics](https://link.springer.com/book/10.1007/0-387-28695-0), a framework which aims to describe all components of a graph. The `ggplot2`-package relies on this framework hence the name. This package is already included in the `tidyverse` therefore you do not have to install it again. If you load the `tidyverse`-library, the `ggplot2`-library is loaded automatically.

<aside>Even art can be created with the `ggplot`-library as shown [hier](https://www.data-imaginist.com/art).</aside>

```{r}
#| warnings: FALSE
#| errors: FALSE
#| output: FALSE
library(tidyverse)

d <- read.csv("data/DatasaurusDozen.csv") |>
    filter(condition %in% c("away", "bullseye", "circle", "dino", "dots, star")) |>
    mutate(id = as.factor(id))

d_summary <- d |> group_by(condition) |>
    summarise(mean_x = mean(x),
              mean_y = mean(y))
    
```

<aside>The data we used is from Matejka and Fitzmaurice (2017) and can be found [here](https://dl.acm.org/doi/10.1145/3025453.3025912)</aside>


A graph contains always:

-   ***data***

-   ***geoms***, visible forms (*aesthetics*) such as points, lines or boxes.

-   ***a coordinate systems / mapping*** describes how data and geoms are linked, also colors or grouping variables are specified here

Further components could be:

-   statistical parameters

-   positions

-   coordinate functions

-   ***facets***

-   scales

-   ***themes***

(we will only cover the contents in italics)

::: {.callout-tip appearance="simple"}
## Good to know

-   For plotting with `ggplot()` it is easiest when your data is in *long* format.

-   What variables do you want to plot (categorical? continuous? ...) affects which `geoms`can be used. You can try out what is suited with the `esquisse`-package below or find ideas [here](https://www.data-to-viz.com).
:::

## Data, geoms and mapping

We start with entering the current data frame and add `geoms` and mappings (specified with `aes()`) with arguments such as

```{r}
ggplot(d_summary, # data
       aes(x = mean_x, y = mean_y)) + # mapping
    geom_point() # geom 
```

Depending on your variables and what you want to show with your data different geoms are well suited.

Examples of available geoms:

-   data points, scatterplots: `geom_point()`
-   lines, tendencies: `geom_line()`
-   histograms: `geom_histogram()`
-   means and standard deviations: `geom_pointrange()`
-   densities: `geom_density()`
-   boxplots: `geom_boxplot()`
-   violin plots: `geom_violin()`

<aside>[Here](https://github.com/rstudio/cheatsheets/blob/main/data-visualization.pdf) you can download `ggplot`-Cheatsheet</aside>

## Facets

With facets you can show subsets of your data in different panels

```{r}
ggplot(d, # data
       aes(x = x, y = y)) + # mapping
    geom_point() + # geom 
    facet_grid(~ condition) # facet
```

## Themes and labels

```{r}
ggplot(data = d,
       mapping = aes(x = x,
                     y = y)) +
    geom_point() +
    ggtitle ("Title") +
    labs(title = "Title", 
        x = "Variable A  [a.u.]",
        y = "Variable B [a.u.]") +
    theme_minimal() # also theme_classic and theme_minimal are nice
```

::: {.callout-tip appearance="simple"}
## `esquisse`-package

With this package you can use the data frames in your current environment or load a new one to try out which geoms might be useful

```{r}
#| eval: FALSE

install.packages("esquisse")

esquisse::esquisser() 
```
:::

<aside>If we do not want to load the whole library we can call the function by writing the library name then `::` and the function afterwards such as `esquisse::esquisser()`. This is helpful when libraries have overlapping function names which can create conflicts or to easily call a function without typing `library()` first.</aside>

Further helpful ressources:

-   [`ggplot2`documentation](https://ggplot2.tidyverse.org/)

-   [Website PsyTeachR: Data Skills for reproducible research](https://psyteachr.github.io/reprores-v3)

-   [Start of the PsyTeachR videos by Lisa DeBruine](https://youtu.be/90IdULVGmYY), e.g. [Basic Plots](https://youtu.be/tOFQFPRgZ3M), [Common Plots](https://youtu.be/kKlQupjD__g) and [Plot Themes and Customization](https://youtu.be/6pHuCbOh86s)

-   [Einführung in R](https://methodenlehre.github.io/einfuehrung-in-R/chapters/05-plotting.html) by Andrew Ellis and Boris Mayer

-   [Here](https://www.data-to-viz.com) you find ideas for plots.

-   [Demo of different geoms](https://rstudio-connect.psy.gla.ac.uk/plotdemo)