Interactive visualizations

Static plots allow full control over display, but intuition and understanding are built through interaction.

This can be just as true for your data management tasks as it can be for representing results of analyses.

plotly

Commercial data visualization company, integrating d3 code with python and R.

plotly builds html widgets that can be stand-alone websites, or be embedded into other tools (R-markdown, shiny).

Install and load library

install.packages('plotly')
library(plotly)

basic functionality: ggplotly()

library(ggplot2)

# re-use toy lakes dataset
source('create-sim-lake-data.R')

# basic plot
p <- ggplot(dat, aes(x = lake, y = chlA))
p + geom_boxplot()

# invoke plotly simply by single function - default is to 
display the last ggplot object
ggplotly()

Examine resulting plotly object, noting:

• opens in Viewer window (html code)
• hovering displays values
• can open in browser (it is standalone html)
• can capture a screenshot (png file)

interactive filtering and zooming

p <- ggplot(dat, aes(x = algal_sp, y = chlA, col = lake)) + 
		geom_point()
ggplotly()

Examine resulting plotly object noting:

• ability to filter by clicking on the legend
  • click individual values on/off
  • double-click to isolate values
• ability to zoom using interactive buttons

Do Exercise 3: Add plotly to R-markdown

native interactive plot function plot_ly()

plot_ly(x = dat$algal_sp, y = dat$chlA, 
		type = 'scatter', mode = 'markers', 
        color = dat$lake)

The plot_ly() function encodes many of the idealized design features of good visualizations, but can be more flexible than the ggplot() framework.

See the plot_ly in R cheatsheet