Successful Sabre 5.0 Workshop held on 17 th - 19 th July 2007 at Lancaster University.

A new release of Sabre statistical software was introduced during a recent workshop “An Introduction to Multivariate Multilevel GLMs using SabreR 5.0” in Lancaster. Sabre is statistical software for the analysis of multi-process random effect response data. Sabre software is capable of running on high-performance parallel computers, e.g. those on the UK National Grid Serivce, and outperforms alternative software used for similar statistical analyses; see the comparisons at http://sabre.lancs.ac.uk/comparison3.html.

The workshop demonstrated the new features of Sabre 5.0 (multilevel, multivariate generalised linear models), with over 60 worked examples/ exercises from published work. A new development for Sabre is the SabreR 5.0 plug-in for is a powerful open-source language and environment for statistical computing and graphics. The development of the plug-in makes Sabre 5.0 functionality available through the familiar R interface, which brings accessibility and ease of use benefits to large communities of scientists requiring specialised high performance statistical software. The workshop included sessions on running parallel Sabre 5.0 on the Grid using the GROWL middleware toolkit also from the R interface (for those very large and complex models you always wanted to estimate!).

The development of Sabre and associated software is a major part of the work of the NCeSS node ‘Collaboratory for Quantitative e-Social Science’ which is funded by the ESRC. The associated ReDReSS Web site (ESRC and JISC funded) will host the presentations and supporting material used during the workshop. Sabre, SabreR and GROWL will be available as part of the ESRC e-Infrastructure.

For further information, material and downloads of Sabre and SabreR, see http://sabre.lancs.ac.uk.

For further information about, see http://www.r-project.org.

For GROWL, see http://www.growl.org.uk.

To view the workshop presentations, see http://redress.lancs.ac.uk.

               

CQeSS is a collaborative project to develop and support Quantitative e-Social Science

News article published: 26th July 2007.