Plenary Speakers
Plenary Speakers
Prof Jonathan Crook
Emeritus Professor of Business Economics
University of Edinburgh Business School
https://www.business-school.ed.ac.uk/staff/jonathan-crook
Prof Crook studied Economics at Lancaster and Cardiff. He has been a Visiting Fulbright Postdoctoral Research Scholar, McIntyre School of Commerce, University of Virginia, USA; a Visiting Fellow, University of Warwick, UK, and a Visiting Fellow, European University Institute, Florence. He is a Fellow of Financial Institutions Center, Wharton School, University of Pennsylvania; an External Research Fellow of the Centre for Finance, Credit and Macroeconomics at the University of Nottingham, and has been elected a Fellow of the Royal Society of Edinburgh and made a Fellow of the Academy of Social Sciences. He concentrates on two research areas: modelling of credit risk and operational risk and economics of consumer credit. The former includes survival and multistate modelling, modelling of loss given default, of exposure at default and stress testing. He is particularly interested in using novel predictors, issues concerning variation over time, capital requirements issues and the use of very large datasets. Specifically, his interest is in models for retail credit of all types as well as credit to SMEs and large corporates. His research on the economics of the consumer credit includes the demand (consumption and finance models), the supply of credit and credit constraints using household level data.
Title: Stress testing behavioural and macroeconomic risks in credit portfolios
Large banks are required to stress test their credit portfolios annually under Basel II. Stress testing credit portfolios to macroeconomic shocks at account level involves parameterising a model predicting probability of default or default rates followed by hypothesising specific shocks, or by simulation to derive a value at risk (VaR) or expected shortfall (ES) 12 months into the future. But the probability of default is also correlated with time varying behavioural variables, which in turn are correlated with the macroeconomy. Simulation studies have estimated the VaR when mutually consistent macroeconomic values have been simulated or when behavioural variables have been simulated but not when both are simulated. In this paper we present a method to simulate both behavioural and macroeconomic variables 12 months into the future whilst maintaining the correlation structure between them to derive a more comprehensive simulation methodology to stress test a credit portfolio.
This is joint work with Viani Djeundje
Prof Gareth James
Deputy Dean & E. Morgan Stanley Chair in Business Administration
& Professor of Data Sciences and Operations
University of Southern California
https://www.marshall.usc.edu/personnel/gareth-james
Prof James is the Deputy Dean and E. Morgan Stanley Chair in Business Administration at the USC Marshall School of Business. He is an expert on statistical methodology in the areas of functional data analysis and high dimensional statistics, with particular application to marketing. Professor James has published extensively and is a co-author of one of the leading books in the field. He is an elected Fellow of both the American Statistical Association and the Institute of Mathematical Statistics. Prof James has won several research and teaching awards, including two Deans awards for Research Excellence, three Golden Apple awards for his MBA courses and the Provost’s Mentoring Award. He has also served as an Associate Editor for journals such as JASA, JRSSB and OR.
Title: Irrational Exuberance: Correcting Bias in Probability Estimates
We consider the common setting where one observes probability estimates for a large number of events, such as default risks for numerous bonds. Unfortunately, even with unbiased estimates, selecting events corresponding to the most extreme probabilities can result in systematically underestimating the true level of uncertainty. We develop an empirical Bayes approach “Excess Certainty Adjusted Probabilities” (ECAP), using a variant of Tweedie’s formula, which updates probability estimates to correct for selection bias. ECAP is a flexible non-parametric method, which directly estimates the score function associated with the probability estimates, so it does not need to make any restrictive assumptions about the prior on the true probabilities. We demonstrate that ECAP can provide significant improvements over the original probability estimates.
Dr Ali Joglekar
Adjunct Professor
University of Minnesota
https://agroinformatics.org/people/joglekar/
Dr Joglekar’s current research focuses on the spatially-explicit characterization of farming and its implications on agricultural productivity throughout the developing and developed world. Her current work as a Research Associate for G.E.M.S/International Agro-Informatics Alliance (IAA) initiative involves developing analytical pipelines to enable the interoperability of the “S” (socio-economic) data with the “G.E.M.” (genomic, environmental and management) data within the G.E.M.S. platform. Her work as a Fellow with the International Science and Technology Practice and Policy (InSTePP) center within the Department of Applied Economics involves creating, enhancing and maintaining spatial bio-economic models and datasets and undertaking research to inform practical and policy decisions of governments, food and agribusiness interests, and others.
Title: GEMS: Supporting Data-Driven Agri-Food Innovation from Molecules to Markets
Abstract: TBA
Prof Saralees Nadarajah
Reader in the School of Mathematics
University of Manchester
http://oldwww.ma.man.ac.uk/~saralees/research.html
Prof Nadarajah is a Reader in the Department of Mathematics, University of Manchester, UK. His research interests include climate modeling, extreme value theory, distribution theory, information theory, sampling and experimental designs, and reliability. He is an author/co-author of four books and has over 900 refereed journal papers published or accepted. He has held positions in Florida, California, and Nebraska.
Title: The drastic Under-representation of African Researchers in Africa-related Research
In an ever more connected world one would expect to see collaborative efforts in academia build bridges between nations, continents and peoples. While the internet and digitisation have broken down boundaries and significantly lowered the financial obstacles as well as delays in time that came with international partnerships, collaborations seem to have been strengthened between industrialised nations. In this talk, I will analyse publication data of articles, notes and presentations on that reference the continent of Africa, as well as nations past or present on the continent and investigate the distribution of author affiliation within and outside the continent.
Prof Emmanuel Lesaffre
Professor emeritus Biostatistics
KU Leuven
https://www.kuleuven.be/wieiswie/en/person/00004228
Emmanuel Lesaffre is emeritus Professor at the Leuven Biostatistics and Statistical Bioinformatics Centre (L-BioStat) of the Catholic University of Leuven as well as visiting Professor at the University of Hasselt, both in Belgium and is honorary professor of Erasmus MC at Rotterdam, the Netherlands. He has more than 600 peer-reviewed publications in biostatistics. He was one of the three founding editors of International Statistical Modelling Journal, he has been Associate Editor of Biometrics and of Biostatistics, and statistical advisor of various clinical journals. He is the founding chair of the International Statistical Modelling Society (IWSM), and a former President of the International Society for Clinical Biostatistics (ISCB). He is an ASA and ISI Fellow, and is honorary member of ISCB and IWSM. He (co)-authored nine books, with the latest book entitled: Bayesian Statistics applied to pharmaceutical research, an edited book with editors E. Lesaffre, G. Baio and B. Boulanger, published in Chapman and Hall, 2020.
Title: Incorporation of historical information in the analysis of current data: A review of Bayesian methods with applications in pharmaceutical research
One of the key features of the Bayesian approach is the possibility to formally include prior information into the analysis of current data. This prior information can come from expert knowledge or from historical data or a combination of both. In this talk we focus on how to include prior information from historical data in pharmaceutical research. A naïve approach is to fully incorporate the historical information into the current analysis. It has been recognized that this most of often not a good idea, because it basically comes down to pooling the historical and current data. Additionally, time trends and conditions that differ between historical and current data may prevent to (fully) include the historical information into the current analysis. In this respect, Pocock (1976) has specified some (stringent) criteria in order to take historical information into account. In the last two decades formal procedures have been suggested for the inclusion of historical information, i.e. the power prior, the meta-analytic and the commensurate prior. We review in this talk especially the first two approaches for a single historical study and multiple historical studies. We will also focus on the use of historical controls. Without doubt the use of historical information is becoming increasingly important in pharmaceutical research because it allows to reduce the current sample size. But these methods are also becoming indispensable in e.g. paediatric studies where information from adult studies can be useful to obtain a clearer view on the efficacy and safety of experimental treatments in children and adolescents. Other applications where the use of historical information is important are bridging studies, trials that incorporate real world evidence and trials on medical devices. We discuss the pros and cons of the formal procedures, and we end the talk with the exploration of their frequentist properties. Applications in pharmaceutical research will illustrate the concepts and results.
Dr McElory Hoffmann, CEO of Praelexis, is an internationally published data scientist and a computer science lecturer at the University of Stellenbosch. In 2013 he founded Praelexis with its agile and innovative business model, born as the result of a keen understanding of the predictive analysis needs of governmental stakeholders as well as corporate clients. The integration of theory and praxis lies at the heart of who Praelexis is. Their data crafters are accomplished engineers, mathematicians and scientists, who are much respected in their respective fields of specialisation.
Dr Johan van der Merwe, Data Science Strategist at Praelexis, has a background in both the Social and Economic Sciences. He consulted in multiple industries before joining Praelexis part-time in 2013 and full-time in 2019. His research focuses on Data Strategy and AI Ethics. He facilitates business case discovery, defines business value and enable businesses for Machine Learning and AI. He also manages machine learning projects and enjoys working in this learning organisation where collaboration and innovation is ingrained in the company culture.
Dr McElory Hoffmann, CEO of Praelexis, is an internationally published data scientist and a computer science lecturer at the University of Stellenbosch. In 2013 he founded Praelexis with its agile and innovative business model, born as the result of a keen understanding of the predictive analysis needs of governmental stakeholders as well as corporate clients. The integration of theory and praxis lies at the heart of who Praelexis is. Their data crafters are accomplished engineers, mathematicians and scientists, who are much respected in their respective fields of specialisation.
Dr Johan van der Merwe, Data Science Strategist at Praelexis, has a background in both the Social and Economic Sciences. He consulted in multiple industries before joining Praelexis part-time in 2013 and full-time in 2019. His research focuses on Data Strategy and AI Ethics. He facilitates business case discovery, defines business value and enable businesses for Machine Learning and AI. He also manages machine learning projects and enjoys working in this learning organisation where collaboration and innovation is ingrained in the company culture.
Title: Ethical Machine Learning in Managing a Health Pandemic
Efficient acquisition and processing of information is a key factor in controlling epidemics. Relevant elements of the health status (e.g. infection state) of individuals or groups of individuals, mobility patterns of citizens, or information on social contacts and social networks are examples of such information. Information processing for decision making may include insights from machine learning models for the analysis of the dynamics of an epidemic using and linking such data. However, privacy interests and other ethical issues have to be carefully considered. In this presentation Drs Hoffmann and Van der Merwe will give an overview of the application of machine learning in an interdisciplinary study where Praelexis is the industry partner, as well as the “Ethical User Story” approach they developed to address ethical concerns.