INLA Short Course at UCSC

INLA Short Course at UCSC

Instructors: Håvard Rue (Havard.Rue@math.ntnu.no) and Daniel Simpson (daniel.simpson@math.ntnu.no), of the Department of Mathematical Sciences at the Norwegian University of Science and Technology (NTNU, in Trondheim)
Venue: 1-day short course sponsored by the Department of Applied Mathematics and Statistics at the University of California, Santa Cruz (UCSC), with organizational support provided by the San Francisco chapter of the American Statistical Association (ASA) and the Bay Area chapter of the Association for Computing Machinery.

Location: Room 180 (Simularium) of the Engineering 2 (E2) building (on the ground floor, facing the plaza between E2 and the Baskin Engineering (BE) building) on the UCSC campus; please see here for a campus map (click on "Baskin Engineering Building" in the Facilities pull-down menu at the bottom of the page).

Description: INLA (Integrated Nested Laplace Approximations) is a relatively new method for making fast and accurate approximate Bayesian calculations in latent Gaussian Markov random field models. This extremely useful class includes time series models, generalized additive models (GAMs), generalized additive mixed models (GAMMs) for longitudinal data, geoadditive models, models with nonparametric fitting of interaction terms between continuous predictors, and stochastic volatility models; see this introductory document (written by Håvard Rue) for more details.
Participants in this short course (a) will receive a good grounding in latent Gaussian Markov random field models in general, and in a number of special cases of these models (mentioned above) in particular, and (b) will learn how to use the freeware implementation of INLA, for fitting these models, available here in R.

Schedule:
- 8-8.15am coffee and registration in the plaza between BE and E2
- 8.15-9.15am session 1: Introduction; basic INLA ideas
- 9.15-9.30am break
- 9.30-10.30am session 2: The R-INLA package; simple examples
- 10.30-10.45am coffee/tea/pastry break (in the plaza between BE and E2)
- 10.45-11.45am session 3: The R-INLA package -- advanced features
- 11.45am-noon break
- noon-1pm session 4: Gaussian Markov Random Fields (GMRFs) as spatial models
- 1-2.30pm lunch (in the plaza between BE and E2)
- 2.30-3.30pm session 5: Stochastic Partial Differential Equation (SPDE) models
Cost: $50 (to cover facilities and catering), payable at registration (in the form of a check made out to UC Regents).

Pre-registration: Please send an email message, with INLA Registration in the subject line, to David Draper (draper@ams.ucsc.edu), no later than 9am Fri 12 Apr 2013, to indicate your intention to participate in the course (this is needed to confirm the final headcount for catering).

Parking for off-campus participants: Make your way to the Core West Parking Structure (at the corner of McLaughlin Drive and Heller Drive: see here for a campus map (click on "Core West Parking Structure" in the Facilities pull-down menu at the bottom of the page). At the entrance ramp on Heller Drive, between 7.30am and 8am, someone will be there to give you your parking permit and tell you where to park (identify yourself as a participant in the INLA short course).

Who will benefit from this course: The target audience is anyone with basic training in Bayesian hierarchical models and MCMC (e.g., via the BUGS or JAGS language) who wants to learn about R-INLA: a fast, R-based, non-MCMC alternative to BUGS/JAGS.

For additional information: Please email the local organizer, David Draper (draper@ams.ucsc.edu).

visits since creation on 28 Mar 2013.

(last modified: 8 Apr 2013)