Learning outcomes

Know how to apply mixed and generalized modelling, multivariate statistics, and interpret the output, including that published in scientific papers.

 

Goals

Train the students on applying and interpreting mixed and generalised models, and on multivariate statistics.

 

Content

Module 1 : Linear statistical modeling
Theoretical introduction into mixed and generalized models (6h); Practical sessions on R (14h);  Two case studies on mixed and generalized models (4h+2h).

Module 2 : Multivariate data analysis
This module details how to visualize and check multivariate data, how to summarize and combine a set of continuous variables into a lower number of variables through PCA (Principal Component Analysis), how to perform the PCA equivalent for categorical data(FCA, Factorial Correspondence Analysis), and how to unravel the links between two sets of continuous variables (Canonical Correlation Analysis). The teaching philosophy insists on the fact that statistics are tools and that the key skills the student should acquire is the expertise to choose the right tool for the job, how to parameterize it and interpret its results critically. Real examples from ecology will be used to illustrate clean but also more difficult cases, closer to real life.

 

Table of contents

Module 1 (Linear statistical modelling) :

  • A theoretical introduction to LMM and GLMM
  • Application to data: behaviour of mangrove killifish
  • Application to data: red wood ants

 

Module 2 (Multivariate data analysis) :

  • visualize and check multivariate data
  • PCA (Principal Component Analysis),
  • PCA for categorical data (FCA, Factorial Correspondence Analysis)
  • Canonical Correlation Analysis

 

Exercices

Learn to solve a statistical problem. Find the appropriate analysis when faced with a problem, check the application conditions relating to the use of this analysis, perform the statistical test on the R software, interpret the results obtained and illustrate them.

Other information

A basic knowledge of the R software is required: the student is expected to be able to create and modify R-data sets independently and perform basic data management and statistical analysis procedures. If such knowledge is not acquired, the student must be trained autonomously in these skills, e.g. by means of the many resources available online for free.

Assessment method

The two modules will be evaluated separately, each module contributing 10/20 to the final score. As the final score must be an integer number, the sum of the two notes will be rounded up if both modules are passed (at least 5/10) and down if it is not the case.

Module 1 (Linear statistical modelling) :
Open book exam, including two exercises on LMM and GLM(M) on R (based on practical sessions and first seminar)  and one case study (based on second seminar).
 

Module 2 (Multivariate data analysis) :
Open book written exam consisting of multiple choice questions, open questions and practical solution of exercises with R software on a computer. The exam is carried out on Moodle, in a computer room on campus, unless health regulations require that the exam be taken at a distance.

Sources, references and any support material

All resources are available on the Moodle website: visuals of the lectures and practical sessions, data sets and R scripts, links to additional resources and supporting books. Course visuals and materials supporting the practical work are available on Moodle.

Language of instruction

Français