Giampiero Marra - Professor of Statistics

UCL logo

Giampiero Department of Statistical Science
University College London
Gower Street
London WC1E 6BT

Office: Room 142
1-19 Torrington Place
London WC1E 7HB

Phone: +44 (0) 20 7679 1864
Fax:     +44 (0) 20 3108 3105
E-mail: giampiero.marra -AT- ucl.ac.uk

Profile

I work as a Professor of Statistics at the Department of Statistical Science at University College London. After having graduated in Statistics and Economics at the University of Bologna in 2004, I worked as an econometrician and statistician for a consulting firm and a multinational company. In 2007 I was awarded an MSc in Statistics at UCL and I defended my PhD thesis at the University of Bath in November 2010. I joined UCL in September 2010.


Main research interests

Penalized likelihood based inference in semiparametric simultaneous joint equation models, copula regression, generalized additive modelling, distributional regression, generalised additive models for location, scale and shape, flexible survival modelling. Keywords: endogeneity, non-random sample selection, MNAR missing data, observed and unobserved confounding, penalised regression spline, copula, generalized regression, joint models, computational statistics, gamlss, survival data.

As highlighted by my published papers, I engage considerably in cross-disciplinary work in several areas such as probabilistic risk assessment, transport studies, political science, HIV and cancer research, credit risk, crime and perceived social trust. I am a member of the following groups at UCL: General Theory and Methodology, Computational Statistics, Biostatistics, Stochastic Modelling and Time Series, Statistics for Health Economic Evaluation.

I am interested in taking on PhD students working on methodological and applied topics (see my publication list to get a feel about my work). Feel free to get in touch with me and include a brief CV and research plan (if any).

A list of published papers is here


Editorial service

Associate Editor of Statistics and Computing, 2015 -

Member of Editorial Advisory Board of Dependence Modeling, 2018 -

Associate Editor of Journal of the Royal Statistical Society: Series C, 2018 -

Associate Editor of Statistical Methods & Applications, 2019 -

Associate Editor of AStA Advances in Statistical Analysis, 2020 -


GJRM package R package

For the past 10 year I have been extending, maintaining and developing the freely available GJRM R package which incorporates the algorithms for applying the methods and models proposed in all my research papers. The availability of software implementing theoretical ideas is increasingly gaining importance in data science and allows for transparent and reproducible research as well as faster dissemination of scientific results.

The GJRM package provides functions for fitting several joint regression models with various types of covariate effects (e.g., linear, non-linear, spatial) and responses (discrete, binary, continuous, survival), for several situations (e.g., endogeneity, associated model equations, non-random sample selection, partial observability). All model's parameters can be specified as functions of flexible additive predictors. This package was originally designed to fit semiparametric bivariate probit models (hence the name of the previous version of the package called SemiParBIVProbit) but has enormously expanded since then. The fitting functions available in the package allow researchers to fit:

  • univariate GAMLSS or distributional regression models.
  • survival models (with and without informative censoring).
  • copula regression models with binary responses (where the link functions are not restricted to be probit). This is useful to fit joint binary models in the presence of non-random sample selection or correlated/associated responses or endogeneity or partial observability.
  • joint copula models with continuous/discrete/binary/survival margins in the presence of correlated/associated responses or endogeneity.
  • joint copula sample selection models with continuous/discrete/binary response combined with rejection sampling to impute missing outcomes based on MNAR.
  • trivariate binary models in the presence of non-random sample selection or correlated/associated responses.

  • More models and options will be incorporated from time to time. Some of the methods/models in the package have been developed following specific requests we received from researchers and collaborators. Get in touch if you are interested in a particular feature of the package or you would like to discuss potential developments/data analyses. It would also helpful to know what you use GJRM for, whenever you work in academia or industry.

    Selected invited talks/seminars/workshops/short courses

    02/2020: Estimating HIV Prevalence in the Presence of Missingness not at Random by Copula-Based Regression Additive Models (Warwick Medical School, UK)
    01/2020: Estimating the Effect of Insurance on Doctor Visits by Copula-Based Regression Additive Models (School of Mathematics, Edinburgh)
    05/2019: Generalised Joint Regression Modelling with Application to the Effect of Insurance on Doctor Visits (Cass Business School, London)
    03/2019: Generalised Joint Regression Modelling (Speaker, DAGStat 2019, Munich)
    12/2018: Copula additive regression models with endogenous binary treatment and count response (CFE-ERCIM 2018, University of Pisa)
    12/2017: Joint Generalized Survival Models (CFE-ERCIM 2017, University of London)
    07/2017: A new approach to fitting generalised additive models for location, scale and shape (EMS, Helsinki)
    07/2017: Semiparametric Copula-Based Regression Models (Speaker, 38th Annual Conference of the International Society for Clinical Biostatistics, Vigo, Spain)
    05/2017: A Simultaneous Equation Approach to Estimating HIV Prevalence with Non-Ignorable Missing Responses (University of Southampton, CORMSIS Centre)
    02/2017: A Unified Approach to Estimating Bivariate Copula Additive Models (Technische Universitat Munchen, Germany)
    01/2017: One-Day Short Course on Copula Generalised Additive Models for Location, Scale and Shape with R (3rd BIOSTATNET General Meeting, Santiago de Compostela, Spain)
    10/2016: On the Specification and Estimation of Bivariate Copula-Based Regression Models (Research Center for Statistics - UNIGE, Geneva, Switzerland)
    08/2016: Bivariate Copula Additive Models for Location, Scale and Shape (COMPSTAT 2016, Oviedo, Spain)
    09/2016: Two-Day Workshop on Missing Data with Emphasis on Sample Selection Models (Birkbeck, University of London)
    06/2016: One-Day Workshop on Bivariate Copula Additive Models for Location, Scale and Shape (Barclays Investment Bank, London)
    05/2016: One-Day Workshop on Bivariate Copula Additive Models for Location, Scale and Shape (Ministry of Justice, London)
    01/2016: A Copula-Based Regression Framework to Deal with Non-Compliance in RCTs (UCL Medical School, Royal Free Hospital, UK)
    01/2016: Flexible Bivariate Copula-Based Regression (University of Durham, UK)
    08/2015: Workshops on Missing Data using Selection Models (Speaker, CDC in Atlanta and Department of Global Health and Population, Harvard)
    08/2015: A Unified Modeling Approach to Estimating HIV Prevalence in Sub-Saharan African Countries (Georg-August-Universitat Gottingen, Germany)
    12/2014: Copula Regression Spline Models for Binary Outcomes (CFE-ERCIM 2014, University of Pisa, Italy)
    11/2014: Statistical Analysis in Development Economics (Speaker, Erasmus School of Economics, Rotterdam)
    02/2014: Flexible Binary Response Sample Selection Models with Application to HIV Prevalence Estimation (Imperial College London and MRC Clinical Trials Unit London)
    06/2013: Binary Generalized Extreme Value Additive Modelling and Beyond (CFE-ERCIM 2013, University of London)
    12/2012: Practical Variable Selection for Generalized Models (Computing and Statistics ERCIM 2012, Oviedo, Spain)
    02/2012: Modelling Correlated Binary Responses within a Semiparametric Regression Framework (University of Birmingham)


    PDRA and PhD UCL students

    PDRA (tba), EPSRC project, 2021 -
    Alessia Eletti, PhD project on survival modelling, 2020 -
    Malgorzata Wojtys (PDRA), worked on an EPSRC funded project on semiparametric sample selection models, 2012 - 2013
    Ken Liang, PhD project on nonparametric spatio-temporal quantile regression, 2010 - 2014, completed
    Francesco Donat, PhD project on bivariate response ordered probit modelling, 2012 - 2015, completed
    Panagiota Filippou, EPSRC PhD funded project on joint modelling of endogeneity and non-random sample selection, 2013 - 2017, completed
    Karol Wyszynski, PhD project on sample selection models for count data, 2012 - 2015, completed
    Robson Jose Mariano Machado, PhD project on multistate models, 2014 - 2018, completed
    Robinson Dettoni Hidalgo, PhD sponsored by Chilean Government, 2016 - 2020, completed
    Maria Toomik, PhD sponsored by A*STAR, Statistical analysis for highly structured data, 2014 - 2019, completed
    Valentina Marincioni, PhD, Bartlett School of Environment (Energy and Resources), 2014 - 2019, completed

    Teaching and admin duties

    Since 2016
    Undergraduate admission tutor

    Since 2011/2012
    STATG001: Statistical Models and Data Analysis (1st Term)

    2012/2013
    STATG003: Statistical Computing (shared, 1st Term)

    2011/2012
    MATH7501: Probability and Statistics (2nd Term)

    2010/2011
    STAT2001/3101/D2: Probability and Inference (tutorials, 1st Term)
    MATH7501: Probability and Statistics (2nd Term)



    Employment and Education history


    Last modified: December 2020