R Under Sirens

Research, Students, and Community in Wartime Ukraine

Dariia Mykhailyshyna

Kyiv School of Economics   ·   keynote · useR! 2026

“Science and coding are our anchors to normality. Even during blackouts and air raids, the Ukrainian R community continues to analyze data, build packages, and teach students.”

- a Ukrainian R user, 2025 survey

What this talk is

  • A community, seen through a survey of 93 Ukrainian R users
  • What we build - research that the world relies on
  • What we teach - R in classrooms that lose power mid-lecture
  • How we train each other - community as infrastructure
  • What the war actually costs, in their own words
  • And one ask of you

Who we are

A national community - on a map I had to repair

Map of Ukraine, Crimea included with a dashed 'added by hand' outline, shaded by survey respondents per oblast; Kyiv dominates with 74, respondents across 12 oblasts.

93 Ukrainian R users answered from 12 oblasts.

This is Kyiv-heavy, and not a census.

To show Ukraine correctly, I first had to write code to put Crimea back on the map.

More than half were displaced - and most came back

Relocation due to the full-scale invasion: never moved 38 (41%), moved within Ukraine and returned 24 (26%), went abroad and returned 20 (22%), still displaced within Ukraine 11 (12%).

Students, academics, and industry - often at once

Bar chart of respondent roles: academics 28, bachelor students 26, private sector 19, master students 17, PhD 8, government 7.

What they build with R

Bar chart of uses of R: data science 81, statistics/econometrics 64, machine learning 21, teaching 20, automation 15.

A young, growing, intermediate community

How long they have used R

How long respondents have used R: under 6 months 20, 6-24 months 36, 2-5 years 21, over 5 years 16.

Self-rated level (1 = beginner, 5 = expert)

Self-rated R proficiency 1-5: mostly level 3, a solidly intermediate community.

What we build: research under fire

Sleepless nights, safer roads

Scatter map of nearly 2,000 road accidents across Lviv, 2022-2024, clustered along the road network and around the city centre.

Do night-time air-raid alerts change next-day road safety?

In R: ~2,000 geolocated Lviv accidents (2022-2024), linked to alert timing.

The counter-intuitive result: nights with alerts are followed by fewer accidents the next day.

Mapping de-russification, hromada by hromada

Animated choropleth of Ukraine showing the share of streets with Russian-associated names, by hromada, falling over time.

What does de-russification look like on the ground?

I prepared the data; R maps it: the share of streets with Russian-associated names, hromada by hromada, falling over time as communities rename them.

sf + a time axis + animation - a slow civic process you can watch move.

Built in R, used by the world

  • Humanitarian dashboards: IOM DTM Ukraine · dtm.iom.int/ukraine
  • Language infrastructure: translit.uk on CRAN for Ukrainian transliteration
  • Local open data: KSE Loc Data Hub · github.com/kse-ua
  • Policy modelling: migration scenarios · Zenodo

Examples shared by survey respondents - public links only.

Research that is about the war itself

  • Students analyse the air-raid alerts that interrupt their own classes
  • Forecasting models become part of how people reason under uncertainty
  • After the Kakhovka dam was destroyed, R helped measure land-cover change
  • Shiny apps, statistical OSINT, and outage-alert bots turn analysis into response

Examples shared by survey respondents.

What we teach: students under disruption

Preparing a lecture in a metro shelter

Laptop open to RStudio with Quarto lecture notes, resting on a lap in a Kyiv metro station at night, while people shelter on mats and camping chairs in the background.

A real lecture being written - in RStudio, in Quarto - on a metro platform during a night-time missile attack.

I teach Social Statistics (undergraduate, close to data science) and Statistics and Econometrics 1.1.

The materials still shipped on time.

When the alert sounds, class moves underground

A heavy steel blast door with a wheel valve, opening onto an underground room used as a classroom.

Behind this blast door are underground classrooms.

When an air-raid alert sounds during a lecture, this is where the class continues.

R does not care which room it runs in.

R is a good tool for unstable conditions

  • Free and open - install on any machine, no licence to lose
  • Local RStudio / Positron - runs offline, on a laptop battery; no server to reach
  • Quarto - one source rebuilds the whole document; resume after any interruption
  • Git / GitHub - work synced and backed up, safe from any single dead laptop or blackout

“After demobilization, I will need to re-remember R and other programming languages to fully return to civilian work.”

- a Ukrainian R user, currently serving

Workshops for Ukraine

Community as infrastructure

165 workshops delivered

5,163 participants reached

220,720 EUR raised for Ukrainian causes

  • R Workshops for Ukraine - live online sessions donated by international instructors
  • Recordings reach participants when time zones, alerts, or blackouts get in the way
  • Registration fees support Ukrainian causes including Come Back Alive and Leleka

sites.google.com/view/dariia-mykhailyshyna · learn and support Ukraine at the same time

Solidarity, made concrete

  • Expertise moves one way; funding and solidarity move the other
  • Recordings make the model tolerant of time zones, alerts, and blackouts
  • The format creates repeat contact, not one-off sympathy
  • Anyone in this room could teach, sponsor, or host the next workshop

What the war actually costs

“Did the war affect your ability to work in R?”

Bar chart: 59 of 93 say the war affected their ability to work in R 'not at all', 16 slightly, 10 moderately, 8 strongly.

…but look at what actually gets in the way

Even though 63% said the war affected them “not at all”:

Obstacles to R work: power blackouts 50, psychological stress/focus 41, increased workload 38, unstable internet 23, air-raid disruptions 20, no extra limits 19.

The disruption is routine, not rare

Times in the past month war conditions blocked a planned R task: never 55, 1-2 times 26, 3-5 times 9, more than 5 times 3.

And yet - most kept coding, or coded more

Change in how often respondents use R since the full-scale invasion: much less often 2, slightly less 6, about the same 58, slightly more 9, much more 18.

What keeps us coding

“Use R, not Python!”

“Make ML in R great again.”

“I <3 R”

“you gonna hear me RRRR”

Ukrainian R users, 2025 survey

“R lets us speak about and explain the world in a universal language.”

- a Ukrainian R user, 2025 survey

“Human beings can adapt to everything - even to one of the most cruel wars of all times. We will prevail.”

- a Ukrainian R user, 2025 survey

“Do not tolerate russian scientists just because science is outside of politics. Russia is using science as a tool, as a weapon to spread its ideas and narratives, to wash its reputation and to legitimate its wars.”

- a Ukrainian R user, 2025 survey

“We are still here, we are still coding, and we remain a vital, active part of the global scientific landscape. We don’t just need sympathy; we need professional collaboration and integration.”

- a Ukrainian R user, 2025 survey

One ask of you

What this room can do

1 · Collaborate, don’t just sympathise

Co-author, co-supervise, invite Ukrainian researchers into the project.

2 · Cite and integrate

Use and credit Ukrainian work and packages in global projects.

3 · Support community training

Teach a Workshops for Ukraine session, sponsor a cohort, or share recordings.

4 · Keep the door open

Remote-friendly events, speaking slots, and funding for those who cannot travel.

Synthesised from 21 survey messages to the international R community.

“Supporting Ukraine today safeguards our ability to work in R tomorrow.”

- a Ukrainian R user, 2025 survey

“We are still here. We are still coding.”

- a Ukrainian R user, 2025 survey

Backup / Q&A

Backup: Survey method

  • Convenience sample of Ukrainian R users reachable through my network
  • n = 93, Kyiv-heavy; this is a community portrait, not a population estimate
  • Multi-select answers counted with fixed-substring matching, because some labels contain commas
  • Data cleaning, recoding, and figures are reproducible from the project files

Backup: Why Crimea on the map matters

  • Natural Earth is widely used, including by people in this room
  • In the source data I used, Crimea and Sevastopol were not returned as part of Ukraine
  • I repaired the map for this talk because political assumptions can enter technical defaults
  • The point is not cartography trivia; it is infrastructure shaping what becomes visible

Backup: What can I do this week?

  • Email one Ukrainian author whose work you actually use
  • Add a Ukrainian package, dataset, or paper to a syllabus or project
  • Offer one Workshops for Ukraine session, recording, or sponsorship
  • Make one event more remote-tolerant: recordings, asynchronous materials, travel support