Modelos lineales generalizados

Course title

Generalized Linear Models

Faculty

Marta Pérez (Departament d’Estadística i Investigació Operativa, UPC)

Jordi Valero (Escola Superior d’Agricultura de Barcelona, UPC)

Course language

Spanish

Course schedule

Junio 29 a Julio 1: de 9:00am a 1:00pm
Julio 2: de 9:00am a 12:00

Description

Generalized Linear Models (GLMs) are a class of regression models that extend classical liner regression to other types of outcome variables, such as binary or count variables and positive continuous variables. The models may be fitted using the glm function in R. The course includes theory and methodology for GLMs, as well as practical applications using R, where the students use their own computer.

Program

1. Day 1 (4 hours)

   1. Linear regression models
   2. Logistic regression for binomial data

2. Day 2 (4 hours)

   1. Natural exponential families
   2. Poisson regression for discrete data

3. Day 3 (4 hours)

   1. Quasi-likelihood functions
   2. Regression for overdispersed discrete data

4. Day 4 (3 hours)

   1. Gamma regression for positive continuous data
   2. Discussion and perspectives

Evaluation

The evaluation is by practical exercises, where the student writes a short R script to analyze a given set of data. It is complemented by a concepts questionnaire, to be passed in the last session.

Classroom

PC1