CRM > English > Activities > Curs 2015-2016 > Flexible Regression and Smoothing
Flexible Regression and Smoothing
From November 16 to 18, 2015

IMPORTANT NOTE FOR PARTICIPANTS (Updated on November 9th)

Course's description

In all fields of science and technology the amount of data are growing rapidly. Analysing them can be more challenging than ever. Two important issues in any statistical model are the choice of a proper statistical distribution for the response variable and explaining how the parameters of this distribution are changing over different values of explanatory variables. The Generalised Additive Models for Location Scale Shape, (GAMLSS), provides the right framework for that.

GAMLSS is a regression tool box appropriate for medium to large data sets where the distribution of the response variable is allowed to be a very flexible parametric distribution and where all the parameters of the distribution (not only the mean) can be modded using linear or smooth functions of the explanatory variables. GAMLSS allows flexibility in univariate statistical modelling far beyond other currently available methods.

This short course will be an exposition of the GAMLSS framework using throughout practical examples. In particular the following topics will be covered:

An introduction to GAMLSS and its statistical modelling philosophy.

An introduction to the R implementation of GAMLSS.

A description of the different distributions which can be used for modelling the response variable, and their properties. This includes: i) continuous (positively or negatively skewed and with high or low kurtosis) (ii) discrete (over-dispersed or zero in inflated) and iii) mixed distributions.  The different additive terms for modelling the parameters of the distribution will be explored including: linear, non-parametric smoothing and random e effects terms.

Exposition to different modelling selection techniques and diagnostics for checking the model adequacy.

Further statistical modelling examples (including centile estimation).

The course is designed for applied statisticians and PhD students in the
eld of social statistics, biostatistics, medical statistics and other related
fields, where the data requires modelling the response variable using a flexible distribution.

Lecturer: Mikis Stasinopoulos, London Metropolitan University

Scientific coordinators

 Anna Espinal Universitat Autònoma de Barcelona Guadalupe Gómez Universitat Politècnica de Catalunya

Registration​

Registration fee to this course only: 150€

Registration fee to this course and the previous one (see information below): 200€​

Registration includes: Documentation package and coffee breaks

Deadline for registration: November 11, 2015

To register to this page's course click the REGISTER BUTTONS AT  THE PAGE TOP AND BOTTOM​
To register to two courses follow the green arrow below

ATENTION: Reduced registration fee of 200€ for participants also registering to
Competing risks: Concepts, methods and software​, October 14 to 16, 2015. By Ronald Geskus, Academic Medical Center
To register at this fee follow the arrow  (includes registrati​on to both courses)​ CLOSED

​Lodging information

For lodging in the area please click here​

For off-campus and family accommodation click here​​ ​​​
Contact information
​