About Me

Register here

Degree in Mathematics, Università degli Studi di Milano, 1990
Doctorate in Mathematics, Università degli Studi di Milano, 1997
Researcher at CNR-IMATI, November 1996-February 2005
Associate Professor in Probability and Mathematical Statistics, Polimi, March 2005-December 2015
Full Professor in Statistics at Politecnico di Milano from January 2016
Cv in Italian; Cv in English

Research

My research interests are mainly focused on Bayesian inference (parametric and nonparametric), with emphasis on modelling and computational aspects. One of the topic I am currently working on is mixture models where location parameters are a priori encouraged to be well separated, also incorporating covariate information in the likelihood and in the assignment to mixture components; this yields a trade-off between repulsiveness of locations in the mixtures and attraction among subjects with similar covariates. From the application point of view, my work concerns statistics for Medicine (e.g. recurrent events related to hearth failures, survival after an infarction), Health care management (e.g. home care providers, recurrent blood donations), Biology (e.g. SNP's data) and Engineering (e.g. topic modelling); see the list of my papers.

Research keywords

  • MCMC for Bayesian nonparametrics
  • Mixture Models for Cluster Analysis
  • Random probability measures - Dirichlet processes and generalizations - exchangeability and partial exchangeability
  • Bayesian generalized linear mixed models - Regression models for survival analysis

Open the list of papers

Teaching/Theses

All materials for the courses I teach can be find on Beep (students/people with Polimi credentials only).

A list of topics for thesis (graduate, undergraduate, but also PhD's) is below.

Time series (details in Italian) - Tesi di primo o secondo livello (BIO, MTM)

Analisi statistica di serie storiche del peso giornaliero (in Kg) di pazienti affetti da Systemic Capillary Leak Syndrome (SCLS), una malattia estrememente rara; la SCLS provoca frequenti crisi dovute alle improvvive variazioni di pressione arteriosa, che possono risultare fatali. Sono disponibili anche dati del peso giornaliero di pazienti sani. L'obiettivo è anzitutto trovare un modello per serie storiche che si adatti ai dati dei pazienti malati e successivamente l'identificazione dell'arrivo di una crisi. Inoltre, si vuole capire se c'è differenza tra la variazione di peso giornaliera tra pazienti sani e malati.

Metodi statistici: modelli ARMA, modelli di regressione per serie storiche, Hidden Markov models.


Bayesian nonparametric covariate-driven clustering (LM, PhD thesis)

In a Bayesian nonparametric clustering context, the parameter is the random partition of data labels. Conditionally on the partition, data are assumed iid within each cluster and independent between different clusters. Here, we aim at taking into account possible patterns within available covariates: the explanatory variables should drive the prior knowledge on the random partition, specified through a dependent nonparametric prior. In fact, observations with similar covariates should more likely belong to the same cluster, and covariates enter into the prior on the partition of our data via a similarity function. We want to study properties of the similarity function which guarantee optimal clustering features.

Statistical methods: product partition models with covariates (PPMx) and generalizations.

Bayesian models for the analysis of recurrent events (LM, PhD thesis)

The aim is the prediction of the next gap times between two recurrent events, through Bayesian models. Applications concern the analysis of gap times between (i) two successive blood donations from a dataset from AVIS, Milano, (ii) two successive hospitalizations of hearth failure patients.

Statistical methods: Bayesian models for the anaylis of recurrent event data, either via gap times, or via counts.

How to reach me

Email E-mail address: alessandra.guglielmi@polimi.it

Politecnico di Milano, Department of Mathematics, P.zza Leonardo da Vinci 32, 20133, Milano
        Building 14 ("La Nave"), 2nd floor


Tel.: +39 02 23994641; Fax: +39 02 23994513