I’m Marjan Cugmas — statistician, researcher and university teacher. I work at the intersection of statistical methodology and social science, with a particular weakness for networks, clusters, models and charts that make a complicated idea suddenly click.
.jpg)
I develop and apply statistical methods that reveal structure in complex social data, from clusters and network positions to latent dimensions and regression models. I use these methods to study, e.g., social support and scientific collaboration.
I primarily teach statistical methodology at the University of Ljubljana and spend a suspicious amount of time finding ways to make difficult ideas visually intuitive without sacrificing technical rigour.
R packages, dashboards, automatically generated reports, slides, tattoo sketches, tiny web experiments and photo archives. I like analysis most when it becomes something tangible.
“I have a problem. I am an intellectual, but at the same time I am not very clever.”
— Sue Townsend,
The Secret Diary of Adrian Mole Aged 13¾
University of Ljubljana, Faculty of Social Sciences.
University of Ljubljana, Faculty of Social Sciences.
Research and analytics roles at the University of Ljubljana and the National Institute of Public Health.
University of Ljubljana, Faculty of Social Sciences.
University of Ljubljana, Faculty of Electrical Engineering.
University of Ljubljana, Faculty of Social Sciences.
The topics change, but the basic obsession stays fairly constant: what local processes create the larger structure we observe, and how can we measure that structure well?
Who has access to support, who becomes vulnerable, and how support networks shift under major disruptions such as the COVID-19 pandemic.
Longitudinal patterns of collaboration among researchers, doctoral students and mentors, from individual ties to national and European structures.
Survey methodology, text mining, machine learning and network analysis applied to large-scale online health and counselling data.
Coleman’s macro–micro–macro model, network evolution, blockmodeling and the connection between local mechanisms and large-scale structure.
Developing and evaluating approaches to blockmodeling and partition comparison.
R packages, Shiny dashboards, reproducible reporting and tools that convert statistical work into usable systems.
A selection of research and applied projects. A broader record of ARIS-funded projects and programmes is available in SICRIS, although some collaborative and applied work is not included there.
JACARDI is the Joint Action of the European Union that supports European countries in reducing the burden of cardiovascular diseases and diabetes, at both individual and health system levels.
Visit JACARDI ↗Research on mentoring relationships and the dynamics of scientific collaboration, following how doctoral collaborations develop through time.
Project record ↗Comparison and evaluation of approaches to blockmodeling dynamic networks, including Monte Carlo simulation studies.
Project record ↗An interactive web application for real-time analysis and visualisation of questions and counsellors’ answers in the Slovenian online counselling service This is me, including exportable aggregated and micro-level data.
Analysis of two decades of counselling questions, combining hand-coded material with supervised machine learning to reconstruct content labels for earlier years.
Machine-learning identification of cancer-related posts in a large online health community. The best-performing random-forest model reached 92% sensitivity and 89% specificity.
A few recent student projects where methodological guidance became part of the research story.
Timotej Lenič · Master’s thesis
Žiga Rebovšek · Master’s thesis
Eva Gaberšek · Master’s thesis
Vili Smolič · Master’s thesis
Anja Ž. Kunaver · Master’s thesis
A Shiny dashboard built because ordinary weather portals never showed quite the right things — especially how current and forecast weather compare with what is typical.
An 80s-inspired terminal simulator born after a visit to the Slovenian Technical Museum depots in Pivka, created as a one-Sunday experiment with ChatGPT.
Open terminal ↗Custom R code for building the annual Festive Stats Puzzle for Social Informatics students — because the free crossword software was not good enough.
The research gallery brings together conferences, teaching, collaborations, and milestones.
The short films that explain the people, questions and practical value behind the research — without requiring anyone to begin with the methods section.
Large-scale national data reveal different forms of mentor–mentee collaboration and show how mentorship helps early-career researchers grow, gain independence and enter the scientific community.
Marjan CugmasResearch identifying groups of older adults with insufficient support and the implications for care and crisis planning.
Marjan Cugmas · Anuška FerligojA concise introduction to the doctoral programme and the range of specialisations available within statistics.
Marjan Cugmas · Anuška Ferligoj · Bor VratanarRecent teaching spans multivariate analysis, data visualisation and network methodology.
When your dependent variable refuses to behave — binary outcomes, categories, counts or highly asymmetric data — generalized linear models take regression beyond the ordinary linear model.
University of Ljubljana · Applied Statistics MA · 2024–2026How to turn data into visual stories that people can understand and act on — combining analytical thinking, perception, design and storytelling.
University of Ljubljana · Social Informatics MA · 2022–2026Ljubljana Summer School · 2026How to think with data before the methods get complicated — from preparing and exploring data to statistical inference, hypothesis testing and regression.
University of Ljubljana · Social Informatics MA · 2021–2026When there are too many variables to study one at a time — using methods such as cluster analysis, PCA and factor analysis to simplify data and uncover hidden structure.
University of Ljubljana · multiple programmes · 2016–2026Infra4NextGen project · 2025When statistical problems are too complex for analytical solutions — using computational methods to quantify uncertainty, test hypotheses, validate models and handle missing data.
University of Ljubljana · Applied Statistics MA · 2016–2020Hands-on teaching at European network conferences and summer schools.
EUSN + Applied Statistics Conference + SNA Summer School · 2021–2023A practical introduction to working, analysing and thinking with R.
ECPR Summer School · University of Ljubljana · 2015nemBM
Tools for network-evolution models and blockmodeling, including generation of temporal networks under selected local mechanisms.
View on CRAN ↗mri
Measures for comparing partitions defined over overlapping, but not necessarily identical, sets of units.
View on CRAN ↗multiUS
Multivariate-analysis functions developed primarily for teaching applied statistics.
View on CRAN ↗dBlockmodeling
Blockmodeling for signed, valued, one-mode and two-mode networks.
View on CRAN ↗handyReport
Utilities for summary tables and selected statistical tests.
View R-Forge documentation ↗ggthemeUL
ggplot2 helpers for charts aligned with the University of Ljubljana visual identity.
View on CRAN ↗Highlights from a broader publication record, with direct links to the original papers.
Scientometrics · with Anuška Ferligoj, Vladimir Batagelj and Franc Mali.
Scientometrics · with Franc Mali and Luka Kronegger.
Digital Health · content analysis of cancer-related discussions.
Social Networks · with Aleš Žiberna.
Statistical Methods & Applications · with Aleš Žiberna and Anuška Ferligoj.
PLOS ONE · connecting local network mechanisms with global blockmodel structure.
Browse the complete publication archive and open the highlighted conference presentations directly.
I’m into 80s aesthetics, analog photography, old electronics, traditional tattoo art, books, hiking, weather and the strange satisfaction of using an object that has more character than efficiency.
Get in touch about statistical methodology, social networks, research collaboration, teaching or a project idea.