GAMLSS

The GAMLSS working party and other valuable collaborators

Dr. Fernanda De Bastiani

Coauthor of `Flexible Regression and Smoothing:  Using  GAMLSS  in R’ (2017), `Distributions for Location Scale and Shape: Using  GAMLSS  in R (2019), and developer of GAMLSS spatial (Federal University of Pernambuco, Brazil).

Prof. Gillian Heller

Coauthor in three books in GAMLSS, Flexible Regression and Smoothing:  Using  GAMLSS  in R’ (2017). `Distributions for Location Scale and Shape: Using  GAMLSS  in R (2019)  and Generalised Additive Models for Location Scale and shape; a Distributional regression approach with applications (2023) and one in actuarial statistics with Prof. De Jong, Generalized Linear Models for Insurance Data (2008). (NHMRC Clinical trials centre of the University of Sydney, Australia.)

Prof. Thomas Kneib

Leading figure on distributional regression models with several articles on theoretical aspects of Bayesian GAMLSS and GAMLSS with machine learning. Coauthor of "Generalised Additive Model for Location Scale and Shape; A distributional regression approach with Applications."  (Georg-August-Universität Göttingen, Germany.

Prof. Andreas Mayr

Expert in boosting and GAMLSS and  coauthor of "Generalised Additive Model for Location Scale and Shape; A distributional regression approach with Applications,  (Institut für Medizinische Biometrie, Informatik und Epidemiologie, Universität Bonn, Germany)  

Prof. Robert Rigby

Co-creator of GAMLSS, coauthor of `Distributions for Location Scale and Shape: Using  GAMLSS  in R  (2019) and `Flexible Regression and Smoothing: Using  GAMLSS  in R’ (2017). He is the main  developer of distributions and statistical  theory. (School of Computing & Mathematical Science, University of Greenwich, UK).

Prof. Mikis Stasinopoulos

Co-creator of GAMLSS, coauthor of `Flexible Regression and Smoothing: Using  GAMLSS  in R’ (2017), `Distributions for Location Scale and Shape: Using  GAMLSS  in R', (2019) and Generalised Additive Models for Location Scale and shape; a Distributional regression approach with applications (2024). He is the main contributor of the R software. (School of Computing & Mathematical Science, University of Greenwich, UK)

Dr Reto Stauffer

Working on environmental data and expert in R programming, (Department of Statistics and Digital Science Center, University of Innsbruck, Austria).

 

Ass. Prof. Nicolaus Umlauf

Creator of Bayesian GAMLSS (BAMLSS) and expert in R programming, (Department of Statistics, Faculty of Economics and Statistics
Universität Innsbruck, Austria). 

Dr Achim Zeileis

Member of the R code team, editor in Chief for the Journal of Statistical Software,  coauthor of several R packages, including  distributions3, and author of several books. (Department of Statistics, Universität Innsbruck, Austria)

Dr Nikolaos Georgikopoulos

Development, applications and consultant in finance and banking (risk management)m (Stern School of Business-New York University).

Prof. Paul Eilers

Giving valuable advice on smoothing and statistical modelling in general (Erasmus University, Holland).

Dr Marco Enea

Development and applications of the zero inflated and zero and one adjusted models and discrete distributions (University of Palermo, Italy).

Dr Julian Merder

Environmentalist  tackling ecological questions ranging from molecular up to global scales. Promoter of GAMLSS in environmental science, coauthor of the package gamlss.ggplots. (Biosphere Sciences & Engineering, Carnegie Science, Stanford, USA).  

Dr. Luiz Nakamura

Development and application of continuous distributions,  lecturer and researcher at the Federal University of Lavras (Brazil).

Dr. Konstantinos Pateras
 Co-developer of gamlss.demo and co-developing gamlss.shiny. His main research interest is on rare diseases modelling. (Department of Public and One Health, University of Thessaly)

The GAMLSS working Party

There are several packages using the GAMLSS ideas, including  gamlss, bamlss, gamboostLSS etc. The idea of the creation of the GAMLSS working party come during the 2023 IWSM in Dortmund Germany. The aim of the    GWP is to make;

  1. the packages more compatible,
  2. the use of GAMLSS neater for future users, and
  3. the maintenance of the packages durable

The founding member are: Fernanda de Bastiani, Gillian Heller, Thomas Kneib,  Andy Mayr,  Robert Rigby, Mikis Stasinopoulos, Reto Stauffer, Niki Umlauf and Achim Zeileis.    

Contributors

The original GAMLSS software implementation was done in GLIM by Mikis Stasinopoulos and Bob Rigby. The translation from GLIM to R was done in the early 2002 by Mikis Stasinopoulos, Bob Rigby and Popi Akanziliotou. For the more current versions the following people have voluntarily contributed to the GAMLSS software:

  • Elaine Borghie contributed in the improvement of functions centiles.pred(), centiles.split() and Q.stats()
  • Paul Eilers contributed in the creation of the functions scattersmooth(), pb(), pbc() and pvc() and in the package gamlss.demo
  • Steve Ellison contributed to the centiles() function
  • Michael Hohle corrected the function gamlssNP()
  • Larisa Kosidou contributed in the improvement of the package gamlss.demo
  • Brian Marx contributed to the package gamlss.demo
  • Nicoleta Mortan contributed into the creation of the package gamlss.cens
  • Raydonal Ospina contributed the BEOI and BEZI distributions for package gamlss.dist
  • Konstantinos Pateras contributed by creating the nice interfaces in the package gamlss.demo

The following people whose function(s) have been adapted for the GAMLSS software:

  • Gareth Amber for his fractional polynomial function which the gamlss fp() function is based.
  • John Chambers and Trevor Hastie for their R function step.gam() on which the gamlss function stepGAIC.CH() is based
  • Jochen Einbeck, Ross Darnell and John Hinde for their functions alldist() and allvc() from the package npmlreg on which the gamlss.mx function gamlssNP() is based
  • Trevor Hastie for the function random()
  • Jim Lindsey and Philippe Lambert for their function stableglm() in the package stable on which the gamlss.nl function nlgamlss() is based
  • Brian Ripley for his function nnet() to which the gamlss function nn() provides an interface and for the function multinom() which is used in our function gamlssMX()
  • Stefan van Buuren for his original worm plot function on which the gamlss wp() function is based.
  • Venables and Ripley for their R function stepAIC() on which the gamlss function stepGAIC.VR() is based.
  • Simon Wood for his function gam() to which the gamlss function ga() provides an interface.

 

Acknowledgement

We also like to thank the following people who have contributed by suggesting changes or by reporting bugs:

  • Christian Kiffner for suggesting changes for the function term.plot()
  • Albert Wong for suggesting changes to the function lpred() applied to interval response variables.
  • Tim Cole for several suggestions for improvement of the gamlss software.
  • Huiqi Pan for several suggestions to improve centile estimation in gamlss.
  • Willem Vervoort for reporting the problems for calling gamlss() within other functions.