Installation | Syntax | Examples | Feedback | Change log

---

alluvial v1.5

(06 Mar 2025)

Installation

The package can be installed via SSC or GitHub. The GitHub version, might be more recent due to bug fixes, feature updates etc, and may contain syntax improvements and changes in default values. See version numbers below. Eventually the GitHub version is published on SSC.

SSC (v1.42):

ssc install alluvial, replace

GitHub (v1.5):

net install alluvial, from("https://raw.githubusercontent.com/asjadnaqvi/stata-alluvial/main/installation/") replace

The palettes package is required to run this command:

ssc install palettes, replace ssc install colrspace, replace

Even if you have these packages installed, please check for updates: ado update, update.

If you want to make a clean figure, then it is advisable to load a clean scheme. These are several available and I personally use the following:

ssc install schemepack, replace set scheme white_tableau

You can also push the scheme directly into the graph using the scheme(schemename) option. See the help file for details or the example below.

I also prefer narrow fonts in figures with long labels. You can change this as follows:

stata graph set window fontface "Arial Narrow"

Syntax

The syntax for the latest version is as follows:

stata alluvial varlist [if] [in] [weight], [ value(var) palette(str) colorby(layer|level) smooth(1-8) gap(num) recenter(mid|bot|top) shares percent labangle(str) labsize(str) labposition(str) labcolor(str) labgap(str) catangle(str) catsize(str) catposition(str) catcolor(str) catgap(str) valsize(str) valcondition(num) format(str) valgap(str) novalues showtotal novalleft novalright lwidth(str) lcolor(str) alpha(num) offset(num) boxwidth(str) wraplabel(num) wrapcat(num) valprop labprop valscale(num) labscale(num) n(num) * ]

See the help file help alluvial for details.

The most basic use is as follows:

alluvial varlist

where varlist are categorival variables given at the unit of observation. If any variable has more than 10 categories, or it is continuous, the command will throw an error. This is to avoid over-crowding the figure. For suggestions on how to automate this, please open an issue!

Examples

Load a Stata dataset:

stata sysuse nlsw88.dta, clear

Let's test the alluvial command:

stata alluvial race married collgrad smsa union

Smooth

stata alluvial race married collgrad smsa union, smooth(1)

stata alluvial race married collgrad smsa union, smooth(8)

colors

stata alluvial race married collgrad smsa union, colorby(layer)

stata alluvial race married collgrad smsa union, palette(carto)

stata alluvial race married collgrad smsa union, palette(CET I2)

showmiss + shares

stata alluvial race married collgrad smsa union, shares

stata alluvial race married collgrad smsa union, showmiss shares

stata alluvial race married collgrad smsa union, showmiss shares colorby(layer)

gap

stata alluvial race married collgrad smsa union, gap(0)

stata alluvial race married collgrad smsa union, gap(10)

all together

```stata local vars race married collgrad smsa union

alluvial vars', smooth(8) alpha(60) palette(CET C7) gap(10) valcond(100) valsize(2) showtot /// xsize(2) ysize(1) lc(black) lw(0.1) ``

Offset and label rotation (v1.1)

```stata local vars race married collgrad smsa union

alluvial vars', smooth(8) alpha(60) palette(CET C7) gap(10) valcond(100) valsize(2) showtot /// xsize(2) ysize(1) lc(black) lw(0.1) /// laba(0) labpos(3) noval offset(6) ``

Box width (v1.2)

```stata local vars race married collgrad smsa union

alluvial vars', smooth(8) alpha(60) palette(CET C7) gap(10) valcond(100) valsize(2) showtot /// xsize(2) ysize(1) lc(black) lw(0.1) /// laba(0) labpos(3) noval offset(6) boxwid(6) ``

v1.4 features

stata alluvial race married collgrad smsa union, smooth(8) alpha(60) palette(CET C6) valsize(2) /// lw(0.1) lc(black) /// laba(0) gap(2) novalright showtotal /// xsize(2) ysize(1)

stata alluvial race married collgrad smsa union, smooth(8) alpha(60) palette(CET C6) valsize(2) /// lw(0.1) lc(black) /// laba(0) labs(1.6) boxw(10) gap(2) novalues /// showtotal wrapcat(20) wraplab(15) catgap(8) plotregion(margin(b+5 l+10 r+10)) /// xsize(2) ysize(1)

stata alluvial race married collgrad smsa union, smooth(8) alpha(60) palette(CET C6) valsize(2) /// laba(0) labs(1.6) boxw(11) gap(2) novalues /// showtotal wraplab(15) catgap(8) plotregion(margin(b+5 l+10 r+10)) /// xsize(2) ysize(1) labprop showmiss

v1.5 options: percent vs percent2

While percent converts the values to percentages proportional to column heights percent2 shows a pure columnwise share.

stata alluvial race married collgrad smsa union, smooth(8) alpha(60) palette(CET C6) valsize(2) /// laba(0) labs(1.6) boxw(11) gap(2) novalues /// showtotal wraplab(15) catgap(8) plotregion(margin(b+5 l+10 r+10)) /// xsize(2) ysize(1) labprop percent

stata alluvial race married collgrad smsa union, smooth(8) alpha(60) palette(CET C6) valsize(2) /// laba(0) labs(1.6) boxw(11) gap(2) novalues /// showtotal wraplab(15) catgap(8) plotregion(margin(b+5 l+10 r+10)) /// xsize(2) ysize(1) labprop percent2

v1.41

stata alluvial race married collgrad smsa union, smooth(8) alpha(60) palette(CET C6) valsize(2) /// laba(0) labs(1.6) boxw(11) gap(2) novalues /// showtotal wrapcat(20) wraplab(15) catgap(8) plotregion(margin(b+5 l+10 r+10)) /// xsize(2) ysize(1) labprop showmiss percent

stata alluvial race married collgrad smsa union [w = wage], smooth(8) alpha(60) palette(CET C6) valsize(2) /// laba(0) labs(1.6) boxw(11) gap(2) novalues /// showtotal wrapcat(20) wraplab(15) catgap(8) plotregion(margin(b+5 l+10 r+10)) /// xsize(2) ysize(1) showmiss labprop percent

stata alluvial race married collgrad smsa union, value(wage) smooth(8) alpha(60) palette(CET C6) valsize(2) /// laba(0) labs(1.6) boxw(11) gap(2) novalues /// showtotal wrapcat(20) wraplab(15) catgap(8) plotregion(margin(b+5 l+10 r+10)) /// xsize(2) ysize(1) showmiss labprop percent

Feedback

Please open an issue to report errors, feature enhancements, and/or other requests.

Change log

v1.5 (27 Apr 2025) - Complete rework of the base code. This command should now run much faster. - Added option percent2. This shows percentages based on column totals. - catgap is now a percentage of column height to make it consistent across actual values and percentages. - Several bug fixes.

v1.42 (06 Mar 2025) - The program was dropping missing values. This has been fixed.

v1.41 (11 Nov 2024) - Added value() to allow users to specify a numeric variable that can be aggregated over the categories. - Fixed a bug where string variable checks were not working properly. - Minor improvements to the code.

v1.4 (26 Sep 2024) - Added wraplabel() and wrapcat() options that allow users to wrap long labels after specific characters. - Added valprop and labprop to scale labels proportional to their values. The scaling can be adjusted using labscale() and valscale() options. - Added novalleft and novalright to hide left (outgoing) and right (incoming) labels respectively. - Added percent which now shows percentages based on column (layer) height. Note that shares is still available but will scale to [0,1] range. - Added option n() to control how many points are used for calculating the sigmoid curves. - All generic twoway graph options can now be passed onto alluvial such as plotregion(), graphregion() and many others for more control. - Improvements to how weights are used in the command. - Major code cleanup resulting in major speed improvements.

v1.3 (10 Feb 2024) - Options added to control category variables including size, color, gap, angle, position. - Option labcolor() added. - Options graphregion() and plotregion() added. - Minor code cleanups.

v1.21 (19 Oct 2023) - showmiss was not generating the missing values category (reported by Matthias Schonlau). This has been fixed.

v1.2 (04 Apr 2023) - if/in added back in the command. - boxwidth() added to the command. - Minor bug fixes.

v1.1 (15 Jan 2023) - Variable labels are now correctly passing to the final graph (requested by Marc Kaulisch and Ana Karen Díaz Méndez). - Weights option added. Still in beta so more testing is required (requested by Ana Karen Díaz Méndez). - offset option added to extend the x-axis (requested by Marc Kaulisch). - valcond is now just a numeric. It is assumed that the condition implies >= <valcond>. - The missing category, enabled using the showmiss option, now has its own color (requested by Marc Kaulisch).

v1.0 (08 Dec 2022) - Public release.