One command. One file. No server, no Python, no R.
---
📦 Free, open source, one-line install
Requires Stata 17+. Uses Stata's bundled Java - nothing else to install.
GitHub:
https://github.com/fahad-mirza/sparkta_stata
* Please report any bugs
---
Example Code:
That one line produces a fully interactive HTML chart and opens your browser to display a live CI band that recomputes as you drag the slider. No Python environment. No R packages. No Tableau license. Just Stata doing what it already does, and now doing it interactively using Java.
For researchers at institutions with restricted internet or air-gapped systems, there's a built-in "offline" mode. Entire chart library is bundled inside the jar, no CDN calls, no network dependency, nothing to whitelist with IT.
---
📊 What Sparkta does
You write a Stata command. Sparkta builds a fully self-contained .html file that:
→ Opens in any browser, anywhere
→ Has live filter dropdowns and dual-handle range sliders
→ Shows N, mean, SD, median, CV, sparklines all updating live
→ Works offline (air-gapped institutions, reproducibility archives)
→ Can be emailed - the interactivity travels with the file
No installation on the recipient's end. No CDN. No dependencies.
---
📈 This is the updated release and this is what is new in this version
Here's what changed:
FIT LINES + CI BANDS (brand new)
7 types: lfit · qfit · lowess · exp · log · power · moving average
Add fitci and get an asymmetric CI band. Hook it to a slider and the band recomputes live as you drag.
FILTER ENGINE (rebuilt from scratch)
Was: max 2 dropdown filters, pre-computed, HTML grew exponentially
Now: unlimited filters + dual-handle sliders, row-level engine in the browser, O(N) aggregation
REFERENCE ANNOTATIONS (19 new options)
yline() · xline() · yband() · xband() · apoint() · aellipse() · alabelpos()
Draw reference lines, shaded bands, point markers, text labels, ellipses — all offline-safe
COLOR PALETTES (14 themes)
Tableau 10 · ColorBrewer Set1 & Dark2 · Okabe-Ito (colorblind-safe) · viridis · neon
Combine with dark/light backgrounds: theme(dark_viridis), theme(light_cblind1)
PERFORMANCE
500k obs × 100 groups: from 4 seconds → to 0.5 seconds
HTML file: from 40 MB → to 2 MB with the help of lazy sparkline rendering
---
âš¡ Before vs After
Filter variables: before a max of 2, now unlimited
Fit lines: before none, now 7 types
CI bands: before none, now recomputes with sliders
Annotations: before none, now multiple options
Color themes: before 2, now 17
Stats panel: before, static summary stats but now, it updates with filter, and sliders
Performance: previously O(N×G), now O(N)
---
🎯 Who is this for?
Researchers who want publication-quality interactive charts inside their Stata workflow without leaving Stata, learning JavaScript, or setting up a server.
Particularly useful for:
• Conference presentations (self-contained file, works anywhere)
• Supplementary materials (offline HTML = permanent reproducible snapshot)
• Air-gapped institutional networks (offline mode, no CDN)
• Sharing results with non-Stata colleagues
---
Happy to answer questions, drop them in the comments.
Additional Details:
I am trying to get this package submitted to SSC but having trouble sharing files via email to KitBaum due to mail service security reasons as the ZIP contains a JAR file. Hopefully there is some alternate way possible to submit.
Also, special thanks to wbuchanan for introducing this idea in one of his presentations back in 2016 (if i am not wrong).
---
📦 Free, open source, one-line install
Code:
net install sparkta, from("https://raw.githubusercontent.com/fahad-mirza/sparkta_stata/main/ado/") replace
GitHub:
https://github.com/fahad-mirza/sparkta_stata
* Please report any bugs
---
Example Code:
Code:
sysuse auto, clear sparkta price mpg, type(scatter) fit(lfit) fitci sliders(mpg)
For researchers at institutions with restricted internet or air-gapped systems, there's a built-in "offline" mode. Entire chart library is bundled inside the jar, no CDN calls, no network dependency, nothing to whitelist with IT.
---
📊 What Sparkta does
You write a Stata command. Sparkta builds a fully self-contained .html file that:
→ Opens in any browser, anywhere
→ Has live filter dropdowns and dual-handle range sliders
→ Shows N, mean, SD, median, CV, sparklines all updating live
→ Works offline (air-gapped institutions, reproducibility archives)
→ Can be emailed - the interactivity travels with the file
No installation on the recipient's end. No CDN. No dependencies.
---
📈 This is the updated release and this is what is new in this version
Here's what changed:
FIT LINES + CI BANDS (brand new)
7 types: lfit · qfit · lowess · exp · log · power · moving average
Add fitci and get an asymmetric CI band. Hook it to a slider and the band recomputes live as you drag.
FILTER ENGINE (rebuilt from scratch)
Was: max 2 dropdown filters, pre-computed, HTML grew exponentially
Now: unlimited filters + dual-handle sliders, row-level engine in the browser, O(N) aggregation
REFERENCE ANNOTATIONS (19 new options)
yline() · xline() · yband() · xband() · apoint() · aellipse() · alabelpos()
Draw reference lines, shaded bands, point markers, text labels, ellipses — all offline-safe
COLOR PALETTES (14 themes)
Tableau 10 · ColorBrewer Set1 & Dark2 · Okabe-Ito (colorblind-safe) · viridis · neon
Combine with dark/light backgrounds: theme(dark_viridis), theme(light_cblind1)
PERFORMANCE
500k obs × 100 groups: from 4 seconds → to 0.5 seconds
HTML file: from 40 MB → to 2 MB with the help of lazy sparkline rendering
---
âš¡ Before vs After
Filter variables: before a max of 2, now unlimited
Fit lines: before none, now 7 types
CI bands: before none, now recomputes with sliders
Annotations: before none, now multiple options
Color themes: before 2, now 17
Stats panel: before, static summary stats but now, it updates with filter, and sliders
Performance: previously O(N×G), now O(N)
---
🎯 Who is this for?
Researchers who want publication-quality interactive charts inside their Stata workflow without leaving Stata, learning JavaScript, or setting up a server.
Particularly useful for:
• Conference presentations (self-contained file, works anywhere)
• Supplementary materials (offline HTML = permanent reproducible snapshot)
• Air-gapped institutional networks (offline mode, no CDN)
• Sharing results with non-Stata colleagues
---
Happy to answer questions, drop them in the comments.
Additional Details:
I am trying to get this package submitted to SSC but having trouble sharing files via email to KitBaum due to mail service security reasons as the ZIP contains a JAR file. Hopefully there is some alternate way possible to submit.
Also, special thanks to wbuchanan for introducing this idea in one of his presentations back in 2016 (if i am not wrong).
Comment