ggplotline

# ggplotline
## telling a story with labels, colors, and layout
### Malcolm Barrett
### 10/08/2018

<h4>
Slides: <a href='https://malco.io/slides/ggplotline'>malco.io/slides/ggplotline</a>
</h4>

---

## **Don't use too many aesthetics and labels. Be selective.**

---

## ~~Don't use too many aesthetics and labels. Be selective.~~
## **Use color to focus the reader's attention.**

---

## ~~Don't use too many aesthetics and labels. Be selective.~~
## ~~Use color to focus the reader's attention.~~
## **Combine plots from simpler to more complex. Be consistent but not boring.**

---

```r
library(tidyverse)

*scatterplot_extras <- function(legend.position = "none") {
 list(
* theme_minimal(base_size = 14),
 theme(
 legend.position = legend.position,
* panel.grid.minor.x = element_blank(),
* panel.grid.minor.y = element_blank()
 ),
 labs(
 x = "log(GDP per capita)",
 y = "life expectancy"
 ),
 scale_color_manual(values = country_colors)
 )
}
```

---

```r
library(gapminder)

gapminder_2007 <- gapminder %>%
 filter(year == 2007)

gapminder_2007 %>% 
* ggplot(aes(log(gdpPercap), lifeExp, col = country)) +
    geom_point(size = 3.5, alpha = .9) +
    scatterplot_extras("right")
```

---

---

# Direct labeling

1. Label data directly   
2. Ditch the legend
3. Use proximity and similarity (e.g. same color)

---

background-image: url(http://hexb.in/hexagons/ggrepel.png)
background-position: 90% 52%

# ggrepel: Repel overlapping text

```r
library(ggrepel)
```

## `geom_text_repel()`
 
## `geom_label_repel()`

---

```r
gapminder_2007 %>% 
  ggplot(aes(log(gdpPercap), lifeExp)) +
    geom_point(
       size = 3.5, 
       alpha = .9, 
       shape = 21, 
       col = "white", 
       fill = "#0162B2"
     ) +
*   geom_text_repel(
*     aes(label = country),
*     size = 4.5,
*     point.padding = .2,
*     box.padding = .4,
*     min.segment.length = 0
*     ) +
    scatterplot_extras()
```

---

---

## Sample labels rather than display them all

```r
set.seed(1010)

countries <- gapminder_2007 %>% 
* sample_n(15) %>%
 pull(country)

countries
```

```
##  [1] Malaysia              Comoros               Colombia             
##  [4] Nigeria               Taiwan                Sierra Leone         
##  [7] Netherlands           Turkey                Guinea               
## [10] Romania               Sudan                 Sao Tome and Principe
## [13] Tanzania              Korea, Rep.           Albania              
## 142 Levels: Afghanistan Albania Algeria Angola Argentina ... Zimbabwe
```

---

```r
gapminder_2007 %>% 
* mutate(label = ifelse(
*   country %in% countries,
*   as.character(country), ""
*  )) %>%
  ggplot(aes(log(gdpPercap), lifeExp)) +
    geom_point(
      size = 3.5, 
      alpha = .9, 
      shape = 21, 
      col = "white", 
      fill = "#0162B2"
    ) +
    geom_text_repel(
*     aes(label = label),
      size = 4.5,
      point.padding = .2, 
      box.padding = .4, 
      min.segment.length = 0) +
    scatterplot_extras()
```

---

---

# Direct labeling

1. Label data directly   
2. Ditch the legend
3. Use proximity and similarity (e.g. same color)

---

```r
*continent_data <- gapminder %>% 
* group_by(continent, year) %>% 
* summarise(lifeExp = mean(lifeExp))

line_plot_extras <- function(legend.position = "none", 
 values = continent_colors) {
 list(
 theme_minimal(base_size = 14),
 theme(
 legend.position = legend.position,
 panel.grid.major.x = element_blank(),
 panel.grid.minor.x = element_blank(),
 panel.grid.minor.y = element_blank()
 ),
 scale_color_manual(values = values),
 labs(y = "life expectancy")
 )
}
```

---

## Change in average life expectancy by continent

```r
continent_data %>% 
  ggplot(aes(year, lifeExp, col = continent)) +
    geom_line(size = 1.2) +
    line_plot_extras("right")
```

---

---

## Change in average life expectancy by continent

```r
direct_labels <- continent_data %>% 
 group_by(continent) %>%
 summarize(
 x = max(year), 
 y = max(lifeExp)
 )

direct_labels
```

```
## # A tibble: 5 x 3
## continent x y
## <fct> <dbl> <dbl>
## 1 Africa 2007 54.8
## 2 Americas 2007 73.6
## 3 Asia 2007 70.7
## 4 Europe 2007 77.6
## 5 Oceania 2007 80.7
```

---

# cowplot: Manipulate ggplots

```r
library(cowplot)
```

## Themes
## `plot_grid()`
## Manipulating ggplots

---

```r
*p <- continent_data %>%
* ggplot(aes(year, lifeExp, col = continent)) +
* geom_line(size = 1.2) +
* line_plot_extras() +
* scale_x_continuous(expand = expand_scale(0))

direct_labels_axis <- axis_canvas(p, axis = "y") + 
 geom_text(
 data = direct_labels, 
 aes(y = y, label = continent), 
 x = 0.06, 
 hjust = 0, 
 size = 5, 
 col = continent_colors
 )

p_direct_labels <- insert_yaxis_grob(p, direct_labels_axis)

ggdraw(p_direct_labels)
```

---

```r
p <- continent_data %>% 
 ggplot(aes(year, lifeExp, col = continent)) +
 geom_line(size = 1.2) +
 line_plot_extras() + 
 scale_x_continuous(expand = expand_scale(0))

*direct_labels_axis <- axis_canvas(p, axis = "y") +
 geom_text(
 data = direct_labels, 
 aes(y = y, label = continent), 
 x = 0.06, 
 hjust = 0, 
 size = 5, 
 col = continent_colors
 )

p_direct_labels <- insert_yaxis_grob(p, direct_labels_axis)

ggdraw(p_direct_labels)
```

---

```r
p <- continent_data %>% 
 ggplot(aes(year, lifeExp, col = continent)) +
 geom_line(size = 1.2) +
 line_plot_extras() + 
 scale_x_continuous(expand = expand_scale(0))

direct_labels_axis <- axis_canvas(p, axis = "y") + 
* geom_text(
* data = direct_labels,
* aes(y = y, label = continent),
* x = 0.06,
* hjust = 0,
* size = 5,
* col = continent_colors
* )

p_direct_labels <- insert_yaxis_grob(p, direct_labels_axis)

ggdraw(p_direct_labels)
```

---

```r
p <- continent_data %>% 
 ggplot(aes(year, lifeExp, col = continent)) +
 geom_line(size = 1.2) +
 line_plot_extras() + 
 scale_x_continuous(expand = expand_scale(0))

direct_labels_axis <- axis_canvas(p, axis = "y") +
 geom_text(
 data = direct_labels, 
 aes(y = y, label = continent), 
 x = 0.06, 
 hjust = 0, 
 size = 5, 
 col = continent_colors
 )

*p_direct_labels <- insert_yaxis_grob(p, direct_labels_axis)

ggdraw(p_direct_labels)
```

---

```r
p <- continent_data %>% 
 ggplot(aes(year, lifeExp, col = continent)) +
 geom_line(size = 1.2) +
 line_plot_extras() + 
 scale_x_continuous(expand = expand_scale(0))

direct_labels_axis <- axis_canvas(p, axis = "y") +
 geom_text(
 data = direct_labels, 
 aes(y = y, label = continent), 
 x = 0.06, 
 hjust = 0, 
 size = 5, 
 col = continent_colors
 )

p_direct_labels <- insert_yaxis_grob(p, direct_labels_axis)

*ggdraw(p_direct_labels)
```

---

---

# Use color to focus attention

# 1 2 3 4 5 6 7 8 9

---

# Use color to focus attention

# 1 2 3 4 5 6 7 8 9

# 1 2 3 4 5 6 7 8 9

---

---

```r
africa <- gapminder %>% 
 filter(continent == "Africa")

africa
```

```
## # A tibble: 624 x 6
## country continent year lifeExp pop
## <fct> <fct> <int> <dbl> <int>
## 1 Algeria Africa 1952 43.1 9.28e6
## 2 Algeria Africa 1957 45.7 1.03e7
## 3 Algeria Africa 1962 48.3 1.10e7
## 4 Algeria Africa 1967 51.4 1.28e7
## 5 Algeria Africa 1972 54.5 1.48e7
## 6 Algeria Africa 1977 58.0 1.72e7
## 7 Algeria Africa 1982 61.4 2.00e7
## 8 Algeria Africa 1987 65.8 2.33e7
## 9 Algeria Africa 1992 67.7 2.63e7
## 10 Algeria Africa 1997 69.2 2.91e7
## # ... with 614 more rows, and 1 more variable:
## # gdpPercap <dbl>
```

---

```r
africa %>% 
  ggplot(aes(year, lifeExp, col = country)) +
    geom_line(size = 1.2, alpha = .9) +
    line_plot_extras(values = country_colors)
```

---

# gghighlight: Highlight geoms

```r
library(gghighlight)
```

## `gghighlight(predicate)`
## Works with points, lines, and histograms
## Facets well

---

```r
africa %>% 
 ggplot(aes(year, lifeExp, col = country)) +
 geom_line(size = 1.2, alpha = .9) + 
* gghighlight(max(lifeExp) < 50, label_key = country) +
 line_plot_extras(values = country_colors)
```

---

```r
africa %>% 
 ggplot(aes(year, lifeExp, col = country)) +
 geom_line(size = 1.2, alpha = .9) + 
 gghighlight(max(lifeExp) < 50, label_key = country) +
 line_plot_extras(values = country_colors)
```

---

## Which countries had higher life expectancy in 1992 than 2007?

```r
africa <- africa %>% 
 select(country, year, lifeExp) %>% 
* spread(year, lifeExp) %>%
* mutate(le_dropped = `1992` > `2007`) %>%
 select(country, le_dropped) %>% 
 left_join(africa, by = "country")
```

---

## Which countries had higher life expectancy in 1992 than 2007?

```r
africa %>% 
  ggplot(aes(year, lifeExp, col = country)) +
    geom_line(
      size = 1.2, 
      alpha = .9
     ) + 
*   gghighlight(
*     le_dropped,
*     use_group_by = FALSE,
*     label_key = country
*   ) +
    line_plot_extras(values = country_colors)
```

---

---

```r
africa %>% 
  ggplot(aes(year, lifeExp, col = country)) +
    geom_line(
      size = 1.2, 
      alpha = .9, 
*     col = "#E58C23"
    ) + 
    gghighlight(
      le_dropped, 
      use_group_by = FALSE, 
      label_key = labels,  
*     unhighlighted_colour = "grey90"
     ) +
    line_plot_extras(values = country_colors) +
    xlim(1950, 2015) + 
*   facet_wrap(~country)
```

---

---

# Combine plots to tell a story

1. Build plots up from simpler to more complex
2. Don't use the same type of plot in each
3. Use consistent color

---

background-image: url(https://raw.githubusercontent.com/thomasp85/patchwork/master/man/figures/logo.png)
background-position: 90% 45%

# patchwork: Compose ggplots

```r
library(patchwork)
```

## combine plots horizontally: `+`
## combine plots vertically: `/`
## group plots: `()`
## control layout: `plot_layout()`

---

```r
diabetes
```

```
## # A tibble: 403 x 19
## id chol stab.glu hdl ratio glyhb location
## <int> <int> <int> <int> <dbl> <dbl> <chr> 
## 1 1000 203 82 56 3.60 4.31 Bucking…
## 2 1001 165 97 24 6.90 4.44 Bucking…
## 3 1002 228 92 37 6.20 4.64 Bucking…
## 4 1003 78 93 12 6.5 4.63 Bucking…
## 5 1005 249 90 28 8.90 7.72 Bucking…
## 6 1008 248 94 69 3.60 4.81 Bucking…
## 7 1011 195 92 41 4.80 4.84 Bucking…
## 8 1015 227 75 44 5.20 3.94 Bucking…
## 9 1016 177 87 49 3.60 4.84 Bucking…
## 10 1022 263 89 40 6.60 5.78 Bucking…
## # ... with 393 more rows, and 12 more variables:
## # age <int>, sex <chr>, height <int>,
## # weight <int>, frame <chr>, bp.1s <int>, …
```

--
## Assess the relationship between sex, a1c, waist to hip ratio, and body frame

---

```r
label_frames <- function(lbl) paste(lbl, "\nframe")

theme_multiplot <- function(base_size = 14, ...) {
 theme_minimal(base_size = base_size, ...) %+replace%
 theme(
 panel.grid.major.x = element_blank(),
 panel.grid.minor.x = element_blank(),
 panel.grid.minor.y = element_blank(),
 legend.position = "none"
 )
}
```

---

# Combine plots to tell a story

1. Build plots up from simpler to more complex
2. Don't use the same type of plot in each
3. Use consistent color

---

## density plot

```r
plot_a <- diabetes %>% 
 ggplot(aes(waist/hip, fill = sex)) + 
 geom_density(
 col = "white", 
 alpha = .7, 
 size = .75
 ) + 
 theme_multiplot() +
 scale_fill_manual(values = c("#009E73", "#0072B2")) + 
 annotate(
 "text", 
 x = c(.75, .98), 
 y = c(5, 5.70), 
 label = c("female", "male"), 
 color = c("#009E73", "#0072B2"),
 size = 5
 ) + 
* labs(tag = "A")
```

---

---

## boxplot

```r
plot_b <- diabetes %>% 
 drop_na(frame) %>% 
 ggplot(aes(fct_rev(frame), waist/hip, fill = sex, col = sex)) + 
 geom_boxplot(
 outlier.color = NA, 
 alpha = .8, 
 width = .5
 ) +
 theme_multiplot() %+replace%
 theme(axis.title.x = element_blank()) + 
 scale_x_discrete(labels = label_frames) + 
 scale_color_manual(values = c("#009E73", "#0072B2")) + 
 scale_fill_manual(values = c("#009E7370", "#0072B270")) + 
 labs(tag = "B")
```

---

---

## scatter plot with regression lines

```r
plot_c <- diabetes %>% 
 drop_na(frame) %>% 
 ggplot(aes(waist/hip, glyhb, col = sex)) +
 geom_point(
 shape = 21, 
 col = "white", 
 fill = "grey80", 
 size = 2.5
 ) + 
 geom_smooth(
 method = "lm", 
 se = FALSE, 
 size = 1.3
 ) + 
 theme_minimal(base_size = 14) +
 theme(
 legend.position = c(1, 1.25),
 legend.justification = c(1, 0),
 legend.direction = "horizontal",
 panel.grid.minor.x = element_blank(),
 panel.grid.minor.y = element_blank()
 ) + 
 facet_wrap(~fct_rev(frame), labeller = as_labeller(label_frames)) +
 labs(tag = "C", y = "hemoglobin a1c") + 
 scale_y_log10(breaks = c(3.5, 5.0, 7.0, 10.0, 14.0)) +
 scale_color_manual(name = "", values = c("#009E73FF", "#0072B2FF")) +
 guides(color = guide_legend(override.aes = list(size = 5)))
```

---

---

```r
*(plot_a + plot_b) / plot_c
```

---

```r
(plot_a + plot_b) / plot_c
```

---

## Combining patchwork and cowplot

```r
*legend <- ggdraw() +
* get_legend(plot_c + theme(legend.position = "bottom"))

(plot_a + plot_b) / 
(plot_c + theme(legend.position = "none")) / 
legend + 
plot_layout(heights = c(10, 10, 1)) +
plot_annotation(
  "The relationship between waist/hip ratio in males and females by
    frame size", 
  theme = theme(plot.title = element_text(size = 16, face = "bold"))
)
```

---

## Combining patchwork and cowplot

```r
legend <- ggdraw() + 
 get_legend(plot_c + theme(legend.position = "bottom"))

*(plot_a + plot_b) /
*(plot_c + theme(legend.position = "none")) /
*legend +
plot_layout(heights = c(10, 10, 1)) +
plot_annotation(
  "The relationship between waist/hip ratio in males and females by
    frame size", 
  theme = theme(plot.title = element_text(size = 16, face = "bold"))
)
```

---

## Combining patchwork and cowplot

```r
legend <- ggdraw() + 
 get_legend(plot_c + theme(legend.position = "bottom"))

(plot_a + plot_b) / 
(plot_c + theme(legend.position = "none")) / 
legend + 
plot_layout(heights = c(10, 10, 1)) +
*plot_annotation(
* "The relationship between waist/hip ratio in males and females by
*   frame size",
* theme = theme(plot.title = element_text(size = 16, face = "bold"))
*) 
```

---

---

## **Don't use too many aesthetics and labels. Be selective.**

---

## ~~Don't use too many aesthetics and labels. Be selective.~~
## **Use color to focus the reader's attention.**

---

## ~~Don't use too many aesthetics and labels. Be selective.~~
## ~~Use color to focus the reader's attention.~~
## **Combine plots from simpler to more complex. Be consistent but not boring.**

---

# Book Recommendations
## [Fundamentals of Data Visualization](https://serialmentor.com/dataviz/) by Claus O. Wilke
## [Storytelling with Data](http://www.storytellingwithdata.com/) by Cole Nussbaumer Knaflic 
## [Better Presentations](https://policyviz.com/better-presentations/) by Jonathan Schwabish

---

![](https://media.giphy.com/media/3oz8xIsloV7zOmt81G/giphy.gif)

### [malcolmbarrett](https://github.com/malcolmbarrett/)
### [@malco_barrett](https://twitter.com/malco_barrett)

Slides created via the R package [xaringan](https://github.com/yihui/xaringan).

Slides: <a href='https://malco.io/slides/ggplotline'>malco.io/slides/ggplotline</a>