We are happy to announce a seminar by Prof David Warton and Dr Houying Zhu. We hope to see you there.
Any questions, please feel free to contact: secretary.nswbranch@statsoc.org.au.
Date: Thursday, 16th May 2024
Time:
11.00 - 12.00: Houying Zhu
12.00 - 13.00: Complementary lunch with speakers (dietary requirements)
13.00 - 14.00: Prof David Warton
Venue: 12 WW Room 803, Macquarie University and via Zoom.
RSVP: Register at https://www.statsoc.org.au/event-5700496.
Speaker: Prof David Warton (UNSW)
based on work with Elliot Dovers and Jakub Stoklosa
Title: Multivariate Spatial Models for Community Ecology Using a Basis Function Approach
Abstract:
Basis function expansions are a fundamental tool for doing spatial statistics on large datasets or complicated problems. Essentially, a basis function approximation to a random field lets us treat a spatial model like a generalised additive model. Here we will look at three problems from ecology that require the fitting of spatial models to point event or multivariate data and show how we can develop powerful new tools using a basis function approximation, extended (where needed) to the multivariate setting by combining with factor analytical techniques. We show how we can then: fit a log-Gaussian Cox process using standard GAM software; partition biodiversity along spatial gradients to identify regions of high turnover and its main drivers; fit multivariate point process models to account for observer bias and study co-occurrence from citizen science data.
Biography:
David trained to postgraduate level in both statistics and ecology, and his research continues to be at the interface between these disciplines. He has been influential in modernising approaches used in ecology to analyse allometric, multivariate and presence-only data. He advocates a model-based approach where we identify the key properties of data and develop a statistical model to capture these properties, an approach which sometimes necessitates the development of new methodology. He recently wrote a text for ecologists on this theme, published through Springer. David is a Statistics Professor at UNSW Sydney, where he co-founded Stats Central, the university's first statistical consulting unit, which now has eleven staff. He also co-founded the Statistical Consulting Network, which has active members across Australia and New Zealand.
David's research has been honoured with awards from the Australian Academy of Science, the American Statistical Association, and the Royal Society of New South Wales. He is currently interested in developing fast tools for large spatial and spatio-temporal datasets, to better understand ecological communities using presence-only data.
Speaker: Dr Houying Zhu (Macquarie)
Title: Statistical Solutions: From Wildfire Losses to Protein Engineering
Abstract:
This presentation explores two projects utilising statistical methodologies to address diverse challenges. The first project utilises flexible Generalised Additive Models for Location, Scale, and Shape (GAMLSS) to analyse wildfire property losses in the United States. Through the integration of climate, socioeconomic, and geographical variables, our approach enhances catastrophe risk management for insurance companies and government agencies. In the subsequent segment, at a glance, we showcase some fundamental techniques used in assessing a proposed human-centred platform to augment the capabilities of Large Language Models (LLMs) in tackling protein engineering tasks. This collaborative endeavour demonstrates how statistical insights, combined with domain knowledge, empower informed decision-making.
Biography:
Dr. Houying Zhu (https://researchers.mq.edu.au/en/persons/houying-zhu) currently holds the position of Lecturer in Statistics at the School of Mathematical and Physical Sciences at Macquarie University. She earned her PhD in Applied Mathematics from the University of New South Wales, Sydney. Her research is centred on computational statistics, where she pioneers methodologies and fosters efficient implementations. Dr. Zhu's ongoing research interests are geared towards advancing both theoretical understanding and practical applications in computational statistics and data science.
@StatSocAus
