Search

Dr. Gavin Shaddick

Month

December 2015

Short course in Chile – Detecting patterns in space and time: statistical analysis of big data in the power industry

Pca

This short course will be presented by Gavin Shaddick and Amelia Jobling in Santiago on January 12th and 13th. It will cover both theory and applied examples, the latter specifically through practical ‘hands-on’ computer sessions in which participants will be guided through the analyses of real  data.

A guide to the contents is as follows. For more information, please contact Joaquin Fontbona (fontbona@dim.uchile.cl) or Gavin Shaddick (g.shaddick@bath.ac.uk)

The course webpage can be found here: http://www.stat.ubc.ca/~gavin/STEPIBookNewStyle/course_cmm.html

Day 1 AM
Session one: Introduction
Examples of using big data in the power industry including modelling demand patterns, forecasting demands, identifying customer profiles, data reduction techniques.

Session two: Identifying patterns in data and profiling
Clustering and data reduction techniques that allow information to be retained in smaller, more manageable datasets. Examples will include creating profiles for electricity demands for different customer types. Advanced regression models for allocating customers to different demand profiles with measures of uncertainty.

Day 1 PM
Practical sessions using R.
Worked examples of the techniques from Sessions one and two using real datasets.

Day 2 AM
Session three: Spatial modelling
An introduction to Bayesian hierarchical modelling. Exploiting spatial dependence within data to borrow strength. Techniques for lattice and point referenced data. Mapping. Stationary and non-stationary models. Examples will include modelling spatio-temporal modelling of meteorological factors for demand forecasting and others.

Session four: Spatial-temporal modelling
Borrowing information over both space and time. Time series modelling Separable and non-separable models.
Day 2 PM
Practical sessions using R.
Worked examples of the techniques from Sessions three and four including visualising the results from spatio-temporal analyses using Google maps and others.

Advertisements

Second BUC workshop planned for February 2016

BUC2: `Big data: thinking globally

BUC2 will take place at UNAM on 22-24th February 2016. It will include a PhD conference, with participants from Bath and UNAM, and  a short-course centred around dealing with big data, both in terms of statistical techniques and the challenges associated with computation.

MattWorld

Instructors: 

Dr Gavin Shaddick
Dr Daniel Simpson
Dr Karim Anaya- Izquierdo,

Department of Mathematical Sciences, University of Bath, United Kingdom

Abstract:

This course provides an introduction to modelling relationships in both space and time with particular focus on fitting complex models to big data. The course will cover both theory and applied examples, the latter specifically through practical ‘hands-on’ computer sessions in which participants will be guided through the analyses of real  data with both temporal and spatial structure.

Outline:

  • The need for spatial and temporal modelling and the role of big data
  • Bayesian computation
  • Regional modelling with big data
  • Scalable spatial modelling
  • Extending models to further dimensions

 

Teaching methods & Course format:

  • A mixture of lectures, worked examples and computer-based practicals.
  • Students will need to bring a laptop to the course with R and R-INLA software installed (details of how to access these materials will be provided on registration)

 

Learning outcomes:

  • An understanding of the basic principles of hierarchical Bayesian modelling
  • An understanding of the need to account for spatial and temporal dependence in environmental research and, more generally, when modelling complex systems
  • Knowledge of how how integrated nested Laplace approximations (INLA) can be used to implement methods for spatial and temporal modelling with high-dimensional data within a Bayesian framework

 

Textbook:

Spatio-Temporal Methods in Environmental Epidemiology

Gavin Shaddick and James V. Zidek, Chapman and Hall/CRC

 

For further information

Please contact Gavin Shaddick, g.shaddick@bath.ac.uk

 

First BUC workshop – a great success!

Summary

The first BUC workshop took place in CIMAT, Guanajuato, Mexico in the week beginning Monday 9th November 2015. A team of two academics, one PDRA and 4 PhD students from the University of Bath travelled to CIMAT to deliver a short-course, participate in a PhD conference and to engage with colleagues from UNAM and CIMAT in participating in the first BUC executive meeting and to plan the future of the BUC series. The first workshop was a great success with over thirty students and staff attending the short-course and twelve PhD students presenting their research during the conference. The result of a meeting with colleagues from UNAM (in Mexico City) and the executive meeting (in CIMAT) was enthusiastic buy-in to the overall concept of the BUC series from all parties involved and led to detailed planning of BUC2 (February 2016, UNAM), BUC3 (May 2016, CIMAT) and BUC4 (June 2016, Bath) and advanced planning of BUC5 (Q4 2016, 2016).

Picture1

Figure 1: The short-course held during BUC 1

The PhD conference

Within the conference held on the 11th November, four students from CIMAT and 5 from Bath presented their research. Sessions were chaired by students from CIMAT. The conference was well attended, with almost fifty people in the audience over the day.

 

cimatel

Figure 2: CIMATEL, where Bath students were hosted

The short-course

The short-course was the first in a series of three (BUC1, BUC2 and BUC4) in the field of Statistics and Big Data in Environmental Research. The title was `When populations and hazards collide: modelling exposures and health risks’. It was led by GS from Bath and JZ from the University of British Columbia and comprised of a series of lectures and practical, hands-on, computer labs. SAMBa students DF and MT were heavily involved in the successful delivery of the course. In addition to preparing much of the material, they ran the software training labs.

They were kept very busy, with thirty-two students and staff attending the course and labs in the first two days and over 15 staying for an optional lab on Saturday morning!

satAM

Figure 3: Participants at the end of the Saturday morning lab session

Create a free website or blog at WordPress.com.

Up ↑