Stats: sum and quantile

Intermediate Data Visualization with ggplot2

Rick Scavetta

Founder, Scavetta Academy

Recall from course 1

	Cause of Over-plotting	Solutions
1.	Large datasets	Alpha-blending, hollow circles, point size
2.	Aligned values on a single axis	As above, plus change position
3.	Low-precision data	Position: jitter
4.	Integer data	Position: jitter

Plot counts to overcome over-plotting

	Cause of Over-plotting	Solutions	Here...
1.	Large datasets	Alpha-blending, hollow circles, point size
2.	Aligned values on a single axis	As above, plus change position
3.	Low-precision data	Position: jitter	`geom_count()`
4.	Integer data	Position: jitter	`geom_count()`

Low precision (& integer) data

p <- ggplot(iris, aes(Sepal.Length, 
                      Sepal.Width))

p + geom_point()

Jittering may give a wrong impressions

p + geom_jitter(alpha = 0.5,
                width = 0.1,
                height = 0.1)

geom_count()

p + 
  geom_count()

The geom/stat connection

geom_	stat_
`geom_count()`	`stat_sum()`

stat_sum()

p + 
  stat_sum()

Over-plotting can still be a problem!

ggplot(iris, aes(Sepal.Length,
                 Sepal.Width, 
                 color = Species)) + 
  geom_count(alpha = 0.4)

geom_quantile()

ggplot(iris, aes(Sepal.Length,
                 Sepal.Width, 
                 color = Species)) + 
  geom_count(alpha = 0.4)

Dealing with heteroscedasticity

library(AER)
data(Journals)

p <- ggplot(Journals, 
            aes(log(price/citations), 
                log(subs))) +
  geom_point(alpha = 0.5) +
  labs(...)

p

Using geom_quantiles

p +
  geom_quantile(quantiles = 
                c(0.05, 0.50, 0.95))

The geom/stat connection

geom_	stat_
`geom_count()`	`stat_sum()`
`geom_quantile()`	`stat_quantile()`

Ready for exercises!

Intermediate Data Visualization with ggplot2