Speakers
Managing director of Institute of Crop Science, Biostatistics, University of Hohenheim
Hans-Peter Piepho was appointed Professor of Biostatistics at the University of Hohenheim, Stuttgart, Germany in 2001. He has been working as an applied statistician in agricultural research for more than 30 years. His main interests are related to statistical procedures as needed in plant genetics, plant breeding and cultivar testing. Recent interests include envirotype- and marker-enabled breeding, spatial methods for field trials and experimental design for various applications including two-phase experiments and multi-environment trials. Further areas of interest include network meta-analysis and measure of goodness of fit for generalized linear mixed models.
Large agricultural field trials may display irregular spatial trends that cannot be fully captured by a purely randomization-based analysis. For this reason, paralleling the development of analysis-of-variance procedures for randomized field trials, there is a long history of spatial modeling for field trials, starting with the early work of Papadakis on nearest neighbor analysis, which can be cast in terms of first or second differences among neighboring plot values. This kind of spatial modeling is amenable to a natural extension using splines, as has been demonstrated in recent publications in the field. Here, we consider the P-spline framework, focusing on model options that are easy to implement in linear mixed model packages. Two examples serve to illustrate and evaluate the methods. A key conclusion is that first differences are rather competitive with second differences. A further key observation is that second differences require special attention regarding the representation of the null space of the smooth terms for spatial interaction, and that an unstructured variance–covariance structure is required to ensure invariance to translation and rotation of eigenvectors associated with that null space. We develop a strategy that permits fitting this model with ease, but the approach is more demanding than that needed for fitting models using first differences. Hence, even though in other areas, second differences are very commonly used in the application of P-splines, our conclusion is that with field trials, first differences have advantages for routine use.
Research Chair, Statistics, Queensland University of Technology
Kerrie Mengersen is a Distinguished Professor of Statistics and Director of the Centre for Data Science at QUT, Australia. She is an elected member of the Australian Academy of Sciences and the Australian Academy of Social Sciences, and is a Vice-President of the International Statistical Institute. Kerrie is passionate about developing methods to break open data and reveal insights that can help address critical challenges in health, environment, society and industry. She can be found perched on a stool at the intersection of statistics, machine learning, AI, and technology, and being constantly amazed and challenged by the current and future-promised traffic.
Bayesian methods are now pervasive in applied statistical modelling and analysis. One of the benefits - and challenges - of a Bayesian approach is understanding and formulating priors. In this presentation, I will discuss some of our applied projects in environment and health, and the characterisation of objective, semi-informative and subjective priors. Objective priors will be cast in the context of estimating the probability of presence given a sample comprising only absences and no presences. Semi-informative priors will be addressed through consideration of history matching. Subjective priors will be obtained from citizen science in the context of a Virtual Reef Diver project. The formulation of priors for estimation of mixture models and Hidden Markov models will also be discussed. This research is joint with a range of collaborators will be acknowledged in the presentation.
Executive Dean, Faculty of Science and Engineering, Macquarie University
Lucy Marshall is a Professor of Engineering and Executive Dean of the Faculty of Science and Engineering. She is a water resources engineer, with expertise in hydrologic modeling, environmental model optimization, and quantification of uncertainty in water resources analysis. She has a special interest in understanding how environmental observations can be used to quantify uncertainty in systems undergoing change. Her research has spanned the development of new models in the most heavily instrumented watershed in the United States to making flood predictions in ungauged catchments across Australia.
Lucy completed her BEng (Hons), MEngSc, and PhD at the University of New South Wales (UNSW) in Sydney before moving to Montana State University as an Assistant Professor of Watershed Analysis, where she worked at the interface of engineering and environmental science. She returned to Australia as an Australian Research Council Future Fellow at UNSW, and went on to become the director of the UNSW Water Research Centre. She held multiple leadership positions at UNSW, as the inaugural Associate Dean (Equity and Diversity), Associate Dean (Research), and the academic lead for Athena SWAN. She joined Macquarie as Executive Dean in 2022.
Environmental systems are complex, with many interacting variables (biotic and abiotic) across diverse spatial and temporal scales. Historically, modellers have used scientific knowledge of how these systems function to build tractable modelling frameworks. The resultant models are used today in multiple operational settings including natural resource management, early warning systems for hazards such as floods, or ecosystem restoration. This approach to model building is often favoured because it suggests a level of reliability when models are extrapolated to new conditions or scenarios that haven't been observed previously. Despite this, there is an explosion of interest recently in modelling approaches that capitalise on recent advances in machine learning (ML), Artificial Intelligence (AI) and related technologies. These approaches have shown promise in initial applications to environmental problems, but scepticism remains about their usefulness given their disconnect from the knowledgebase of environmental systems. This leads to a fundamental question: How can we effectively combine these innovative approaches with the established understanding of environmental processes? We demonstrate here the integration of traditional process-based environmental models with emerging modelling technologies, aiming to bolster the predictive power and trustworthiness of environmental models. We demonstrate how new hybrid approaches to ML offer a pathway to refining our understanding of natural processes, enhancing predictions, and addressing pressing environmental challenges with improved accuracy.
Founding Executive, VSNi International
We regret to inform you that Roger will no longer be attending in person. Instead, Dr David Baird will be presenting and include some of Roger's work. We apologise for any inconvenience this may cause.
Roger Payne leads the development of Genstat at VSN, now working part-time after 15 years in the full-time role of VSN's Chief Science and Technology Officer. He has a degree in Mathematics and a PhD in Mathematical Statistics from University of Cambridge, and is a Chartered Statistician of the Royal Statistical Society. Prior to joining VSN, Roger was a statistical consultant and researcher at Rothamsted, becoming their expert on design and analysis of experiments, as well as leader of their statistical computing activities. He originally took over the leadership of Genstat there in 1985 when John Nelder retired. His other statistical interests include generalized and hierarchical generalized linear models, linear mixed models, the study of efficient identification methods (with applications in particular to the identification of yeasts), and ornithological applications.
Part 1: “Genstat, the birder's friend”, will describe how Genstat is used to process and check the over 750,000 bird sightings recorded each year in the London area. These need to be imported from a wide range of sources and interpreted, validated and collated so that they can be summarized by the authors of the London Bird Club's annual reports – see London Natural History Society - London Bird Report. This involves many different aspects, including text mining, matching and comparison, UTM coordinates and grid references, report writing, tabulation, manipulation and, of course, the spreadsheet. There is still scope for statistical analysis too, and many of the counts suffer upper-censoring.
Part 2: The use of Genstat to predict the allocation of damage to the Christchurch earthquakes will be described. This used a large (186,000 cases) K-nearest-neighbour analysis that was tuned by simulation, and whose standard errors were calculated using bootstrapping. The model was then validated by manually assessing 4000 properties and comparing these to their predictions.
Part 3: Fun & games – finally, we will play some of the classic games available in Genstat: fruit machine, Conway’s Game of life and noughts and crosses. Although primarily there for a bit of light-hearted fun, these games also demonstrate interactive procedures, and illustrate how to create “movies” by the means of an animated series of graphs or PNG images.
John Curtin Distinguished Emeritus Professor, Curtin University
Adrian Baddeley is a retired professor of statistics whose main research interests are spatial statistics and statistical computing. He is a graduate of the Australian National University and Cambridge University,has worked at Cambridge, Bath, Yale, UWA, CSIRO, CWI Amsterdam, and Curtin University, and held honorary professorships at Leiden and Aarhus. Adrian Baddeley is a winner of the Pitman and Hannan medals for statistical research, holds an honorary Doctorate of Science from Aalborg University, is a John Curtin Distinguished Emeritus Professor at Curtin University, and is a Fellow of the Australian Academy of Science.
The peninsula and islands of Murujuga (Dampier Peninsula) in Western Australia contain over one million panels of indigenous rock art, in close proximity to an industrial area. Concern about potential damage to the rock art caused by industrial emissions has led to a research programme which is monitoring the condition of the rock art, studying the rock surfaces, and monitoring air quality across the peninsula. This talk will describe the essential role played by spatial statistical methods in designing the study and in identifying potential evidence of the effects of industrial emissions on the rock art.
Principal Managing Consultant Statistician, Data Analysis Australia
Anna Hayes (Munday) is an applied statistician with over 20 years’ experience working as a statistical consultant at Data Analysis Australia. Through managing Data Analysis Australia's consulting team, Anna provides leadership and direction to the company and monitors all projects undertaken by consultants. Anna is an Accredited Statistician of the Statistical Society of Australia, as well as being a past President of the Society’s Western Australian Branch. Anna has a wealth of expertise in statistical application, with particular expertise in statistical analysis, forecasting, survey design and analysis, data management, methodology development, statistical reviews, project management and report writing. Anna’s experience covers a vast array of industries, including environmental, biological and agricultural sciences, energy, transport, workforce planning, the Courts and legal applications.
Data Analysis Australia was engaged to design a statistically sound method to determine operational trigger thresholds to determine whether groundwater levels differ beyond what would be considered normal after the building and filling of a series of solar salt evaporation ponds by our client. Due to limited ‘before’ data and high levels of variability in the measured groundwater levels, we developed a method to detect changes using Auto-Regressive Integrated Moving Average (ARIMA) models that use data from corresponding impact and reference bores to forecast expected groundwater levels at the impact bores and define a trigger to occur when the outcome is not within a certain confidence interval from the forecasts. We determined the appropriate reference bores for each impact bore using a process called Dynamic Time Warping (DTW). I will discuss the development of the statistical methodology from the consulting perspective.