When interactive?

There are three main categories of tasks¹ that interactive graphics can be useful for:

1. Exploratory data analysis
2. Understanding models / algorithms
3. When searching for information quickly without fully specified questions

How interactive?

Enter plotly!

• The plotly package builds on what you already know using ggplot
• All you need to do [mostly] is wrap your ggplot calls in a ggplotly() function.

ggplot + plotly

library(ggplot2)
ggplot(mpg, aes(displ, hwy)) + 
  geom_point()

ggplot + plotly

ggplot(mpg, aes(displ, hwy)) + 
  geom_point()

ggplot + plotly

p <- ggplot(mpg, aes(displ, hwy)) + 
  geom_point()
p

ggplot + plotly

library(plotly)
p <- ggplot(mpg, aes(displ, hwy)) + 
  geom_point()
ggplotly(p)

ggplot + plotly

This works for (almost) any ggplot!!

library(plotly)
p <- ggplot(mpg, aes(displ, hwy)) + 
  geom_point()
ggplotly(p)

Interactive Vis

• Create a ggplot examining the relationship between height and mass using the starwars dataset
• Make this plot interactive using the ggplotly() function

More plotly magic

• Often when doing exploratory data analysis, it can be helpful to examine data points more closely
• the highlight_key() and highlight() functions in plotly can allow you to select certain points
• We can then use the crosstalk and DT packages to look more closely at these points

More plotly magic

m <- highlight_key(mpg)
p <- ggplot(m, aes(displ, hwy)) + geom_point()
gg <- highlight(ggplotly(p), "plotly_selected")
crosstalk::bscols(gg, DT::datatable(m))

More plotly magic

m <- highlight_key(mpg)
p <- ggplot(m, aes(displ, hwy)) + geom_point()
gg <- highlight(ggplotly(p), "plotly_selected")
crosstalk::bscols(gg, DT::datatable(m))

Interactive Vis

• Create an interactive plot of height by mass using the starwars data
• Add the ability to highlight specific points on this plot
• Using crosstalk::bscols() and DT::datatable() link this plot to a data table
• Knit your .Rmd file and select the "outlier" to see details about this data point

Interactive maps

• Another fun package in R is the leaflet package - this allows you to create interactive maps

leaflet

• To get started, call the leaflet() function. Then addTiles()
• By default, leaflet uses OpenStreetMap map tiles

library(leaflet)
leaflet() %>%
  addTiles()

+−

Leaflet | © OpenStreetMap contributors, CC-BY-SA

leaflet

• To get started, call the leaflet() function. Then addTiles()
• You can add markers using addMarkers() with latitude and longitude specified

library(leaflet)
leaflet() %>%
  addTiles() %>%
  addMarkers(lng = -80.2765, lat = 36.1334, popup = "Dr. D'Agostino McGowan's classroom")

+−

Leaflet | © OpenStreetMap contributors, CC-BY-SA

leaflet

• You can pass a data frame to leaflet as well
• Dataset of seismic events off Fiji since 1964

data(quakes)
quakes %>%
  slice(1:20) %>%
  leaflet() %>%
  addTiles() %>%
  addMarkers(~long, ~lat, 
             popup = ~as.character(mag), 
             label = ~as.character(mag))

+−

Leaflet | © OpenStreetMap contributors, CC-BY-SA

leaflet

data(quakes)
quakes %>%
  slice(1:20) %>%
  leaflet() %>%
  addTiles() %>%
  addMarkers(~long, ~lat, 
             popup = ~as.character(mag), 
             label = ~as.character(mag))

+−

Leaflet | © OpenStreetMap contributors, CC-BY-SA

leaflet

data(quakes)
quakes %>%
  slice(1:20) %>%
  mutate(color = ifelse(mag > 5, "red", "yellow")) %>%
  leaflet() %>%
  addTiles() %>%
  addCircleMarkers(~long, ~lat,
             color = ~color,
             label = ~as.character(mag))

+−

Leaflet | © OpenStreetMap contributors, CC-BY-SA

Interactive Vis

• Load the leaflet package
• Load the starbucks.csv data
• Subset the starbucks data to only locations in North Carolina
• Use the leaflet package to plot the starbucks locations
• Label the locations by the store_name variable