graph buble with pie charts

[Mina Sami]

Hello any way on Stata to create a buble graph and each bubble can be dissected into parts like a pie chart?

[Joseph Coveney]

My guess is no, but it's only a guess.

So you want a scatterplot with dozens of little, potentially overlapping, variably sized pie charts? If you are sure that your audience will appreciate digesting so much (not strictly related) information in a single graph, then maybe you can create a complex, layered dot chart as an alternative.

Otherwise, consider multiple separate simpler graphs that present the information in conceptually more bite-sized portions.

[Adam Ross Nelson]

I know this is an old one... I found a few other posts asking similar questions. Decided this one looked like a good place to show a possible template that solves the bubble chart problem. A student this semester asked me for help. In short, here is a first draft.

// Standard set the environment
cls
set more off
clear all

// Lets demonstrate a 'three-dimensional' visualization
sysuse auto

// --- This is a 'four-dimensional' with a 'bubble' element

// Create an ordinal based on price
xtile pbubble = price, nquantiles(3)

// Add a few observations that will support the 'legend'
set obs 80

// Enter 'legend data'
replace weight = 4000 in 75
replace weight = 4350 in 76
replace weight = 4700 in 77

replace weight = 4000 in 78
replace weight = 4350 in 79
replace weight = 4700 in 80

replace mpg = 35 in 75
replace mpg = 35 in 76
replace mpg = 35 in 77

replace mpg = 32.5 in 78
replace mpg = 32.5 in 79
replace mpg = 32.5 in 80

replace foreign = 0 in 75
replace foreign = 0 in 76
replace foreign = 0 in 77

replace foreign = 1 in 78
replace foreign = 1 in 79
replace foreign = 1 in 80

replace pbubble = 1 in 75
replace pbubble = 2 in 76
replace pbubble = 3 in 77

replace pbubble = 1 in 78
replace pbubble = 2 in 79
replace pbubble = 3 in 80

twoway (scatter mpg weight if foreign == 1 & pbubble == 1, msize(small) msymbol(circle_hollow)) ///
       (scatter mpg weight if foreign == 0 & pbubble == 1, mcolor(%50) msize(small) msymbol(circle)) ///
       (scatter mpg weight if foreign == 1 & pbubble == 2, msize(large) msymbol(circle_hollow)) ///
       (scatter mpg weight if foreign == 0 & pbubble == 2, mcolor(%50) msize(large) msymbol(circle)) ///
       (scatter mpg weight if foreign == 1 & pbubble == 3, msize(huge) msymbol(circle_hollow)) ///
       (scatter mpg weight if foreign == 0 & pbubble == 3, mcolor(%50) msize(huge) msymbol(circle)), ///
       text(37 4000 "Low Price") ///
       text(37 4350 "Med Price") ///
       text(37 4700 "High Price") ///
       text(35 3750 "Foreign") ///
       text(32.5 3750 "Domestic") ///
       legend(off) name(bubble)