Scales

class: middle, center, title-slide

.title[
# Scales
]
.subtitle[
## Data Visualization
]
.author[
### Johan Larsson
]
.author[
### Behnaz Pirzamanbein
]
.institute[
### The Department of Statistics, Lund University
]

---

## Scales and Ranges

Choosing appropriate scales can make or break a visualization!

One must be **knowledgeable** about what the data represent.

Sometimes the choice is obvious, other times **tricky** to pick scales appropriately.

.pull-left[
<img src="scales_files/figure-html/unnamed-chunk-1-1.png" width="345.6" style="display: block; margin: auto;" />
]

.pull-right[
<img src="scales_files/figure-html/unnamed-chunk-2-1.png" width="345.6" style="display: block; margin: auto;" />
]

.center[*Number of reported crimes in Sweden (https://www.bra.se).*]

---

## Scales and Limits

It is not always appropriate to include entire range, such as when

- 0 is not a meaningful value for the variable,
- only a small subset of the range is attained in practice, or
- when even small differences matter a lot.

<div class="figure" style="text-align: center">
<img src="scales_files/figure-html/unnamed-chunk-3-1.png" alt="Change in reported crimes between 2019 and 2020 in Sweden (https://bra.se)." width="345.6" /><img src="scales_files/figure-html/unnamed-chunk-3-2.png" alt="Change in reported crimes between 2019 and 2020 in Sweden (https://bra.se)." width="345.6" />
<p class="caption">Change in reported crimes between 2019 and 2020 in Sweden (https://bra.se).</p>
</div>

---

## Scales and Limits when using Color

Choosing limits for color-mapped variables can also be problematic!

<div class="figure" style="text-align: center">
<img src="scales_files/figure-html/unnamed-chunk-4-1.png" alt="Life expectancy in the US." width="720" />
<p class="caption">Life expectancy in the US.</p>
</div>

---

## Scales and Limits when using Color

Choosing limits for color-mapped variables can also be problematic!

<div class="figure" style="text-align: center">
<img src="scales_files/figure-html/unnamed-chunk-5-1.png" alt="Life expectancy in the US with other limits [66, 80]." width="720" />
<p class="caption">Life expectancy in the US with other limits [66, 80].</p>
</div>

---

## Size

When mapping a variable to size, map it to **area** (not height or width).

It usually becomes a bubble chart.

```r
p <- ggplot(mtcars, aes(wt, mpg, size = hp)) + geom_point()
```

.pull-left[

```r
*p + scale_size_area()
```

<img src="scales_files/figure-html/unnamed-chunk-7-1.png" width="345.6" style="display: block; margin: auto;" />
]

.pull-right[

```r
*p + scale_radius()
```

<img src="scales_files/figure-html/unnamed-chunk-8-1.png" width="345.6" style="display: block; margin: auto;" />
]

---

## The Lie Factor

.pull-left[
The lie factor is another concept introduced by Edward Tufte

$$
\text{lie factor} = \frac{\text{effect in visualization}}{\text{size of effect in data}}
$$
- should strive for a lie factor of 1
]

.pull-right[
<div class="figure" style="text-align: center">
<img src="images/liefactor_doctor.jpg" alt="Data mapped to height and width of the doctor image, inducing a lie factor of 2.8." width="90%" />
<p class="caption">Data mapped to height and width of the doctor image, inducing a lie factor of 2.8.</p>
</div>
]