You are here Graduatte College Science Graduate Schools BEE Actuality April 15th-19th, 2016 at UCL:one-week training on Statistical analyses with Generalized Linear Models by G. San martin (CRA-W, Belgium)

Tools

Coming soon...

No event

Job opportunities

Administration

April 15th-19th, 2016 at UCL:one-week training on Statistical analyses with Generalized Linear Models by G. San martin (CRA-W, Belgium)

Added Tuesday 9 February 2016

Dear PhD-students and postdocs in Biodiversity, Ecology and Evolution,

 

The BEE doctoral school organizes a one-week training with Gilles San Martin, from the CRA-W (Gembloux, Belgium), on Statistical analyses with Generalized Linear Models.

This course will take place over 5 half-days of 4 hours, see detailed description below.

The training can be given in French or in English but most the slides and exercises will be in French.

It will take place at UCL (Louvain-la-Neuve, Belgium), every morning on April 25th to 29th, 2015.

The number of participants is limited to 20 and a participation fee of 100 Euros is requested. (In case the participation fee is too high for your budget, please let us know and we will find a solution).

Also, we insisted that you have the necessary prerequisites as listed in the flyer to attend this training - see list below.

In the case you would like to attempt the training, please apply by sending an email to camille.turlure@uclouvain.be, with your name, lab, supervisor, title and brief description of the PhD thesis or research project, and why you want to attend this course before April 1st, 2015.

As the number of participants is limited, selection will be based on the date of application and motivation.

An email of confirmation for the selected persons will be sent on April 10th.

 

Many thanks in advance!

The organisers,

Camille Turlure & Caroline Nieberding


Detailed information on training;
The schedule will be divided into 5 morning sessions for "applied theory". Additional correct-ed exercises will be provided . So please reserve some time during the evenings to do the exercises on your own and assimilate the content of the theoretical sessions. The training can be given in French or in English but most the slides and exercises will be in French.
The "statistical tests" you know (such as stu-dent t-test, anova, regression, G-test,...) are in fact specific cases of Generalized Linear Models (GLMs). GLMs are currently one of the most used approach to analyse univariate ecological data because they are extremely flexible and can be applied to a large diversity of experi-mental designs or observational studies inclu-ding binary data, contingency tables, grouped data, repeated measures, nested designs,... Hence, this training will start from the more simple statistical tests you have heard of and show you how to move to GLMs and exten-sions in R and with real life datasets.
We will insist a lot 1) on the interpretation of the numerical output provided by the software, 2) the building of graphics to help the interpre-tation of the models, 3) the conditions of appli-cations of these models, how to check them and which solution exist when the conditions are not met.
The theory will be kept to the minimum, we will see the theory only when it allows a better applied use of these statistical methods.

Schedule and prerequesite:
Day 1 : Simple Linear Regression, ANOVA1 & Post-Hoc comparisons Day 2 : Multiple Linear Regression & ANCOVA,
Interactions formula notation in R models com-parisons (Type I, type II, type III) Day 3 : Conditions of application (mainly : lineari-ty, normality homogeneity of variances). General-ized Linear Models (binomial, poisson, "log-linear" models, overdispersion) Day 4 : (Generalized Linear) Mixed Models Day 5 : Model selection and multi-model inference GLMs in the real life (conditions of application, independence of the residuals with spatial or tem-poral data, ...)
WARNING You need the following prerequisites:
- R language: data importation and exploration, efficient use of the R help, installation of packages, basic knowledge on the graphic system, loops and conditional execution.
- Basic statistics: Null hypothesis testing, probabil-ity distribution, interpretation and differences be-tween confidence interval, standard error and de-viation, interpretation of a p value, good experi-mental designs.

Last update by EDT BEE on Tuesday 9 February 2016