Generalized Additive Models (GAM) with R

The Finnish Society of Biostatistics (FSB) is organising a 3-day course in Generalized Additive Models.

The course will provide a modern introduction to generalized additive models based on penalized regression spline smoothing. Topics covered will include: types of smooth (including thin plate splines, tensor product smoothing and adaptive smoothing), smoothness selection (REML, GCV, AIC etc), random effects models, model building and checking, interval estimation, testing and prediction. 2 days of lectures will be followed by one day of computer lab work, using R. The course is supported by Tampere Graduate School in Information Science and Engineering (TISE)

The course is given by Simon N Wood (Mathematical Sciences, University of Bath, UK).

Outline

The lectures (21-22.9.2010) will start at 9:00 and end at 17:00. The computer exercises (23.9.2010) will start at 8:00 and end at 16:00. See venue below for details on the lecture rooms.

Day 1, morning staring at 9:00
• Introduction. What is a GAM; some motivating example analyses.
• Revision of background. Linear models, generalized linear models, mixed models and R.

Day 1, afternoon
• Basis-penalty smoothing. The simplest smoother: piecewise linear smoothing; bases and penalties; penalized fitting and smoothing parameters.
• Smoothness estimation. Cross validation and its relatives; the Bayesian smoothing model; REML smoothness estimation.
  
  Seminar ends for the day at 17:00

Course dinner at 19:00 in restaurant Telakka, Tullikamarin aukio 3

Day 2, morning starting at 9:00
• GAMs in general. The model; representation of the model; types of smoother; types of smooth term.
• GAM estimation and checking. The fitting method; checking residuals; checking other assumptions.

Day 2, afternoon
• Further inference. Interval estimation; hypothesis testing; model comparison; prediction; posterior simulation.
• GAMs in practice. Practical use of GAMs with the mgcv package in R.
  
  Seminar ends for the day at 17:00

Day 3, starting at 8:00
This day will consist of practical computer exercises.
• Build you own basis-penalty smoother
• The specify-fit-check-interpret process for GAMs.
• Practical basic analysis.
• Less standard models (e.g. with functional data predictors and random effects).
  
  Seminar ends at 16:00

Registration

The registration deadline is 10th September. To register please go to the registration page.

Participant's fees include
- admittance to the course (21-23.9.2010),
- course materials,
- course dinner (21.9.2010).

NOTE! The early bird registration deadline has been extended to August 22, 2010.

Venue

The course will take place at the University of Tampere. The seminar will take place in Pinni B building in lecture room 3107 (3rd floor), Kanslerinrinne 1. The workshop will be held in the Linna building in computer class 51, Kalevantie 5, in computer room 51. Pinni B and Linna are closely situated, within walking distance from Tampere railway station and bus station (< 1 km) (see map).