Survey responses
The basics: 16 open-ended answers, cleaned, counted and packed so no two words touch.
Stata code
calligram, textvar(response) minfreq(1)Click any picture to open its interactive page.
A title, a light background and a gradient
Colours run from the most to the least frequent word.
Stata code
calligram, textvar(response) minfreq(1) bgcolor(white) ///
gradient(#1d3557 #e63946) title("What customers said")
Rotated words, another font
Seven fonts are bundled, so the page looks the same on every computer.
Stata code
calligram, textvar(response) minfreq(1) rotate(random) font(Roboto)
A heart
Nine shapes are built in: rectangle, ellipse, circle, square, cloud, heart, diamond, triangle and star.
Stata code
use cg_inaugural, clear
calligram, textvar(text) shape(heart) width(1000) height(900) ///
maxwords(250) rotate(random)
A star with its outline
outline draws the edge of the shape behind the words.
Stata code
calligram, textvar(text) shape(star) outline width(1000) height(1000) ///
maxwords(250) font(Lato)
Any shape from coordinates
Points saved as _ID, _X, _Y form a one-area shapefile; the outline is exact.
Stata code
calligram, textvar(text) shapes(cg_blob_shp.dta) projection(none) outline ///
width(1000) height(1000) maxwords(300) phrases rotate(90) font(Roboto)
... or from a picture
Any black-on-white image works as a mask, including a Stata graph you export.
Stata code
graph export cg_mask_blob.png, width(1000) replace
calligram, textvar(text) mask(cg_mask_blob.png) outline ///
width(1000) height(1000) maxwords(300) rotate(random) font(Lato)
Every city inside its own country
A world shapefile and each city's latitude and longitude; size shows population.
Stata code
use cg_cities, clear
calligram, wordvar(city) countvar(population) shapes(cg_countries_shp.dta) ///
location(lat lon) outline outlinewidth(1) outlinecolor(#b8b8b8) ///
width(2400) height(1067) maxwords(1207) minfreq(1) minfontsize(4) ///
margin(2) font(Lato) bgcolor(white)
... and each near its real location
near places every city as close as it can to its own point.
Stata code
calligram, wordvar(city) countvar(population) shapes(cg_countries_shp.dta) ///
location(lat lon) near outline ...
A continent, cut with ordinary Stata commands
keep and merge on the shapefile give any region you like.
Stata code
use cg_countries, clear
keep if continent == "Africa"
keep _ID
merge 1:m _ID using cg_countries_shp, keep(match) nogenerate
save africa_shp, replace
use cg_cities, clear
calligram if continent == "Africa", wordvar(city) countvar(population) ///
shapes(africa_shp.dta) location(lat lon) outline width(900) height(1280)
... with words turned to fill each country
rotate(random) lets city names run along tall or narrow countries, so countries with few cities are less empty.
Stata code
calligram if continent == "Africa", wordvar(city) countvar(population) ///
shapes(africa_shp.dta) location(lat lon) outline rotate(random)
Two-word phrases
phrases finds pairs such as "united states" and "fellow citizens" and shows each as one entry.
Stata code
use cg_inaugural, clear
calligram, textvar(text) phrases maxwords(200) font(Merriweather) ///
title("US inaugural addresses, 1789-2025")
One cloud per group
by() puts each group in its own tab, and image() writes one picture per group.
Stata code
calligram, textvar(text) by(century) phrases maxwords(120) font(Merriweather)
A book in a circle
Alice's Adventures in Wonderland, with "mock turtle", "march hare" and "white rabbit" found as phrases.
Stata code
use cg_alice, clear
calligram, textvar(text) phrases shape(circle) width(1000) height(1000) ///
maxwords(250) minfreq(3) rotate(random) font(Merriweather) ///
bgcolor(#fdfaf3) gradient(#7b2d26 #33658a)
Words already counted
wordvar() and countvar(): car makes in sysuse auto, sized by total price.
Stata code
sysuse auto, clear
gen brand = word(make, 1)
calligram, wordvar(brand) countvar(price) minfreq(1) nostopwords ///
shape(ellipse) font(Open Sans) bgcolor(white)
Survey weights
Every word counts with its response's weight (aweight, fweight).
Stata code
calligram [aweight = weight], textvar(response) minfreq(1)
The results as a dataset
saving() keeps every word with its count, size, position and colour.
Stata code
calligram, textvar(text) phrases maxwords(100) saving(words, replace)
use words, clear
list rank word count fontsize in 1/10
3 million words in under a second
80,000 paragraphs, counted on all processor cores.
Stata code
use cg_inaugural, clear
expand 50
calligram, textvar(text)
Smaller pages with online
Words as text with the font from Google Fonts: much smaller files, needs internet.
Stata code
calligram, textvar(text) online font(Lato) maxwords(200)
Colour by a number
Each word gets the average year of the addresses it appears in.
Stata code
calligram, textvar(text) colorby(year) gradient(#d95f02 #5e3c99) phrases ///
maxwords(200) font(Merriweather)
Colour by group
Each word takes the colour of the century it is most typical of; click the key to hide a group.
Stata code
calligram, textvar(text) colorby(century) phrases maxwords(200) font(Merriweather)
Groups from value labels
A labelled variable gives the categories and their names.
Stata code
sysuse auto, clear
gen brand = word(make, 1)
calligram, wordvar(brand) countvar(price) colorby(foreign) minfreq(1) ///
nostopwords shape(ellipse) font(Open Sans) bgcolor(white)
Colours from a picture
Each word takes the average colour of the picture under its letters.
Stata code
calligram, textvar(text) shape(heart) colorimage(stripes.png) ///
width(1000) height(900) maxwords(250) rotate(random) bgcolor(white)
Every language and writing system
20 languages, 12 writing systems: right-to-left Arabic and Hebrew, Hindi, Tamil, Thai, Chinese, Japanese, Korean. Click the key to hide a script.
Stata code
use cg_languages, clear
calligram [fweight = responses], textvar(text) colorby(script) ///
minfreq(1) maxwords(250) scaling(0.35) font(Lato)
Urdu in Nastaliq
Customer comments in Urdu, in Urdu's own calligraphic style, Nastaliq, with the bundled Noto Nastaliq Urdu font.
Stata code
use cg_urdu, clear
calligram [fweight = responses], textvar(text) minfreq(1) ///
stopwords(اور تھی تھا ہے لیکن مگر بہت ہوئی) ///
scriptfonts(arabic "Noto Nastaliq Urdu") shape(circle)