Associate Professor of Statistics · University of Cagliari
Luca Frigau
I develop statistical learning methods (tree-based models, clustering methods and text mining) and apply them to questions in tourism, health, transport and the life sciences.
Department of Economics and Business Sciences, University of Cagliari, Italy.
Research
Questions I work on
Five lines of research, each told as a story with an illustration: the question it started from and how the work has developed since.
Clustering
How do we find groups in data that we can trust?
Methods to find groups of similar observations, also with the help of an outcome variable, and to validate partitions and choose between clustering methods.
Decision trees and classification
How can a prediction rule be both reliable and easy to read?
From the reliability of classifiers and their use on real problems to oblique regression trees, which split the data with hyperplanes and remain easy to read.
Text mining and topic modeling
What can statistics learn from what people write?
Topic models and sentiment analysis turn reviews, course evaluations and conference programs into data, and can help to schedule a better conference.
Counts and rounded answers
What do rounded answers hide?
Self-reported counts pile up on round numbers. We modeled these spikes, instead of ignoring them, to study the age at which people start smoking.
Statistical image analysis
How can we get reliable information from images?
Images of seeds, from segmentation and classification to validation: PARSEG validates a segmentation with a small sample of its pixels, with roughly the same precision.
News
Recently
- AnalisiMercato, a jamovi module for my Market Analysis course (sampling, hierarchical cluster analysis and principal component analysis, with an Italian interface), is now on GitHub (opens in a new tab).
- At ENBIS 2026 in Florence I organized the ISBIS invited session Advances in Statistical Analysis for Large-Scale Dynamic Networks, Labour Market Outcomes, and Interpretable Ensemble Methods.
- New book chapter on the territorial determinants of tourist arrivals in Italy, based on multivariate regression trees, in the Springer series of the Italian Statistical Society (doi:10.1007/978-3-032-30881-8_7 (opens in a new tab)).
- At IFCS 2026 in Milan I organized the CLADAG invited session Advances in Latent Structure Modeling and co-authored three contributions, on regression trees, topic extraction and ensembles of oblique trees.
Publications
Selected work
Applies GAITD regression and doubly constrained reduced-rank VGLMs to self-reported counts that pile up on round numbers, to study the age at which people start smoking in a large US health survey.
Tree Oblique for Regression with Weighted Support Vector Machine (opens in a new tab)
Regression trees whose splits are oblique hyperplanes found by a weighted support vector machine. In the experiments of the paper, TORS most often had the lowest prediction error among six tree models.
Optimizing the JSM Program (opens in a new tab)
Seeded topic models and a scheduling algorithm reduce the overlap of topics between parallel sessions of the Joint Statistical Meetings.
Network-based Semi Supervised Clustering (opens in a new tab)
Introduces network-based semi-supervised clustering (NeSSC): trees guided by an outcome estimate pairwise affinities, and community detection on the resulting network finds the clusters.
Software
Code, data and apps
Code and data that accompany my articles. Interactive R Shiny applications will be listed here as they are released.
Research code · 2025
PARSEG
R implementation of PARSEG (PArtitioning, Random Selection, Estimation and Generalization), a data-efficient framework for pixel-level image segmentation that selects the smallest number of labeled pixels needed for stable and reliable results. Includes a reproducible example.MIT license.
Data · 2021
Health care fraud classifiers in practice
Repository accompanying the article Health care fraud classifiers in practice (Applied Stochastic Models in Business and Industry, 2021). It provides the labeled payment data set used in the article.
Teaching
I teach Statistics at bachelor’s level and Market Analysis at master’s level at the University of Cagliari, and contribute to doctoral and second-level master’s programs.
Contact
The best way to reach me is by email at frigau@unica.it. Office hours are published on my university page.