Comparison Slider for figures and images with RICV

There is no end to all the magic you can do with R. I come across at least one nifty tool in the web everyday. RICV is one such awesome little visualization package that lets you compare plots using sliders.

Here’s a short introduction of RICV’s magic.

Get set up

Install the package from github and call the libraries you need to create visualizations. ggplot is my viz workhorse. I also call ggthemr because it lets me use the rust theme whose color palettes work nicely with this website theme :).

# devtools::install_github("xvrdm/ricv")
library(dplyr)
library(ggplot2)
library(ricv)
library(ggthemr)

Invoke the theme

Let’s get the rust theme

ggthemr('dust')

Generate a few plots

I got the IRIS dataset to test a few bivariate scatterplot variants. p1 is a basic plot. p2 is a bin2d plot where the rectangle shading usually helps when there is overplotting.

plot_gg <-  ggplot(data=iris, aes(x=Sepal.Width,y=Sepal.Length)) + 
  geom_point() +   xlim(2, 4.5) + ylim(4, 8)

p1 = tempfile(fileext = '.png') %>% 
                basename() %>% 
                  paste0('img/', .)

p2 = tempfile(fileext = '.png') %>% 
                basename() %>% 
                  paste0('img/', .)

ggsave(p1,plot_gg)

ggsave(p2,plot_gg + 
         geom_bin2d()+theme(legend.position = "none"))

Check the plots

They look right. Now let’s combine them!

Combine plots and bring the slider

ppp <- ricv(img1 = p1, img2 = p2,
     options = list(addCircle = TRUE, hoverStart = TRUE))

ppp

## $addCircle
## [1] TRUE
## 
## $hoverStart
## [1] TRUE
## 
## NULL
## $addCircle
## [1] TRUE
## 
## $hoverStart
## [1] TRUE
## 
## attr(,"class")
## [1] "Options"

We now have the plots overlaid and a slider lets us interact and visualize changes or get a different visual perspective of the bivariate relationship.

Generate a regression overlay

Let’s try a variant that is especially useful - regression lines that can be overlaid and swept away on demand when exploring scatterplot. p3 creates a smooth regression line with confidence intervals.

p3 = tempfile(fileext = '.png') %>% 
                basename() %>% 
                  paste0('img/', .)

ggsave(p3,plot_gg + 
         geom_smooth())

Awesomeness

We now can slide regression lines in and out of plots.

ppp1 <- ricv(img1 = p1, img2 = p3,
     options = list(addCircle = TRUE, hoverStart = TRUE))

## $addCircle
## [1] TRUE
## 
## $hoverStart
## [1] TRUE
## 
## NULL
## $addCircle
## [1] TRUE
## 
## $hoverStart
## [1] TRUE
## 
## attr(,"class")
## [1] "Options"

ppp1

Many possibilities

That’s it, folks. Try out RICV to add a bit of literal slickness to your visualizations. I’ll try to use this utility when showing changes over time, compare model estimates, or display maps with different filters.