gghdr: Graphing highest density regions using grammar of graphics

# gghdr: Graphing highest density regions using grammar of graphics
### Sayani Gupta Sayani07     SayaniGupta07 <a href="https://sayanigupta-useR2020.netlify.com/" class="uri">https://sayanigupta-useR2020.netlify.com/</a>
### useR! 2020

---

# Graphing distribution summaries

.left-column[
- several ways to summarize a distribution
- R package ggdist for visualizing distributions and uncertainty
]

# Different summarization expose different features

.left-column[
- Boxplot not suitable for summarizing multi-modal distributions
- often useful to display different summarization in unison
]

background-position: center
background-size: contain

---
# Highest Density Region (HDR) plots

If `$f(x)$` is the density function of a random variable `$X$`, then the 100 `$(1 -\alpha)$` HDR is the subset `$R(f_\alpha)$` of the sample space of `$X$` such that `$R(f_\alpha) = \{x: f(x) \geq f_\alpha\}$`, where `$f_\alpha$` is the largest constant such that `$Pr(X \in R(f_\alpha)) \geq 1-\alpha)$`.

---
# Graphing HDRs using R package hdrcde

Data set: Waiting time between eruptions and the duration of the eruption for the Old Faithful geyser in Yellowstone National Park, Wyoming, USA.

```
## # A tibble: 272 x 2
## eruptions waiting
## <dbl> <dbl>
## 1 3.6 79
## 2 1.8 54
## 3 3.33 74
## 4 2.28 62
## 5 4.53 85
## 6 2.88 55
## 7 4.7 88
## 8 3.6 85
## 9 1.95 51
## 10 4.35 85
## # … with 262 more rows
```

]

---

# Graphing HDRs using R package hdrcde

<img src="images/allplots1png.png" width="90%" />
---
# Extending ggplot2

- HDR is a novel technique for summarizing distribution for which plotting is not implemented in ggplot2
- ggplot2 creates graphics based on The Grammar of Graphics
- graphics are built step by step by adding new elements that allows for extensive flexibility and customization of plots
- need to extend the functionality of ggplot2

---
class: top left

# Package gghdr

- implement HDR plots in ggplot2 
- key elements: *geom* and *stats*
- inspired from R package [`hdrcde`](https://pkg.robjhyndman.com/hdrcde/) developed by [Rob
Hyndman](https://robjhyndman.com/)
]

.animated.bounce[
<img src="images/gghdr_logo.png" height=280px>
]
]

---
# HDR boxplots

![](index_files/figure-html/hdrcde-boxplot-1.png)

]

```r
library(gghdr)
library(ggplot2)
ggplot(faithful, aes(y = eruptions)) +
  geom_hdr_boxplot()
```

![](index_files/figure-html/geom-hdr-boxplot-1.png)
]

---

# HDR rug plots and scatter plots

```r
faithful %>% 
  ggplot(aes(x = waiting, 
             y = eruptions)) + 
  geom_hdr_rug(fill = "blue") + 
  geom_point() 
```

![](index_files/figure-html/hdr-rug-1.png)
]

```r
faithful %>%
 ggplot(aes(x = waiting,
 y = eruptions)) + 
 geom_point(aes(colour = 
 hdr_bin(x = waiting,y = eruptions)))
```
 
![](index_files/figure-html/hdr-scatter-1.png)

]
---
class: top left

# Keep combining - HDR box + jitter

```r
faithful %>% 
 ggplot(aes(y = eruptions)) + 
*geom_hdr_boxplot(fill = c("blue")) + 
*geom_jitter(aes(x = 0))
```
- jitter to supplement the insight drawn from the HDR boxplot

]

---
class: top left

# Keep combining - HDR scatter + HDR marginal

.pull-left[
```r
faithful %>% 
ggplot(aes(x = waiting, y = eruptions)) +
* geom_point(aes(colour = hdr_bin(x = waiting, y = eruptions)))+
* geom_hdr_rug()
```

- Both bivariate and marginal HDRs displayed at once
- Bimodality in both marginal and bivariate distributions

]

![](index_files/figure-html/combo-1.png)
]

---

# Authors

![](images/authors1.png)

---
class: middle top

#  More Information

Package: https://github.com/ropenscilabs/gghdr

Slides: https://sayanigupta-useR2020.netlify.com/

Materials: https://github.com/Sayani07/useR2020

Slides created with Rmarkdown, knitr, xaringan, xaringanthemer