Ihaka Lecture Series

Lecture 1: Thursday 28 July 2022

The genesis of experimentation

Dr Emi Tanaka, Senior Lecturer in Statistics, Monash University

Experiments are essential endeavours to understand the process or phenomena around us via the analysis of experimental data. However, as a precursor to any analysis, the importance of the design of experiment and the data collection process cannot be emphasised enough.

There is no salvation for rubbish data, yet there is far more focus on the analysis of experimental data than any steps prior to the analysis. In this talk, I introduce the framework, called “the grammar of experimental designs”, implemented as the edibble R-package that puts the focus on capturing the user’s intention and understanding of the experimental structure to plan, design and simulate experiments. This approach differs considerably from standard, often recipe-driven, approaches and has potential to encourage users to reflect and revise designs tailored to their experimental need.

Dr. Emi Tanaka is a lecturer in statistics at Monash University whose primary interest is to develop impactful statistical methods and tools that can readily be used by practitioners. Her research area includes data visualisation, mixed models and experimental designs, motivated primarily by problems in bioinformatics and agricultural sciences. She is currently the President of the Statistical Society of Australia Victorian Branch and the recipient of the Distinguished Presenter’s Award from the Statistical Society of Australia for her delivery of a wide-range of R workshops. 

You can see the slides from this lecture here, and watch the lecture on YouTube here.

Lecture 2: Thursday 4 August 2022

New plumbing: Adding a pipe operator to base R

Professor Luke Tierney, Ralph E. Wareham Professor of Mathematical Sciences, the University of Iowa

The forward pipe operator ‘%>%’ was introduced to R by the ‘magrittr’ package and has since become an integral part of many data science workflows. Forward pipe operators have also been introduced in other languages in recent years. R 4.1.0 added a new forward pipe operator ‘|>’ to the base R language. This talk will review the history of forward pipe operators in R and other languages, and explain the motivation and design decisions behind the new operator.

Luke Tierney is Ralph E. Wareham Professor of Mathematical Sciences at the University of Iowa. He has been a member of the R Core Team since 1998. His research has focused mainly on two aspects of computational methods and tools to support statistical analysis. The first area involves developing computational methods, based on approximations and simulation methods, for carrying out Bayesian data analysis. The second involves designing, developing, and maintaining computing environments for statistics and data science.

You can watch the lecture on YouTube here.

Lecture 1: Thursday 29 July 2021

Data Science in the Connected Era

Dr Simon Urbanek Senior Lecturer, Department of Statistics, University of Auckland

Our world is increasingly interconnected, which has several implications. On the one hand it increases the amount and variety of data we can collect to make informed decisions and improve our lives, but also it allows us to perform data analyses without constraints related to the physical location of the data or compute infrastructure.

Modern computer technologies such as cloud computing and the Web have given rise to social media, but in this talk we will explore the possibilities of leveraging them for visualisation and data analysis, connecting people with data across the world and fostering collaboration.

We will illustrate the benefits of that approach using RCloud - a collaborative tool for data analysis and interactive visualisation which supports several data analytic languages, distributed computing, discovery, sharing and reproducible research. It allows us to analyse data collaboratively at a large scale and communicate results efficiently.

Professor Simon Urbanek is a Senior Lecturer in the Department of Statistics at the University of Auckland. Simon obtained his PhD in Statistics from the Augsburg University, Germany in 2004 and has worked at AT&T Labs in Data Science and AI Research for 15 years, leading research and projects on large-scale data analysis in the areas of mobility networks, TV and advertising.

His main interests are visualisation, interactive graphics, big data analytics, statistical and distributed computing. He is member of the R Core Development Team and author of numerous popular R packages including Rserve, multicore, rJava, iPlots, RJDBC and iotools.

Livestream available

Lecture 2: Thursday 5 August 2021

Implementing a Machine-Learning Tool to Support High-Stakes Decisions in Child Welfare: A case study in Human Centred AI

Professor Rhema Vaithianathan, Centre for Social Data Analytics, AUT

Data analytics techniques like predictive risk modelling offer incredible opportunities to learn from rich data sets and make decisions supported by data. But while the private sector has been quick to realise the benefits of data analytics (especially as a tool to drive profitability), the public sector has moved much slower, despite needing new solutions to many wicked social problems.

Professor Rhema Vaithianathan will reflect on what we can learn about applying data analytics in a trusted way, from the very different experiences of the private and public sectors. In particular, she will talk about different approaches to key concepts like consent, transparency, fairness and community voice and how they can contribute to project success or failure. She will go on to talk about new ‘rules of engagement’ that are emerging for social good uses of data analytics, drawing on her experiences implementing the Allegheny Family Screening Tool, a machine learning tool used to support screening of child abuse calls in Allegheny County, PA (United States) since 2016, and scaling out of this work in California and Colorado.

Professor Vaithianathan is a Professor of Economics at Auckland University of Technology where she is director of the Centre for Social Data Analytics, a research centre focused on using data analytics for social impact. She is also a Professor of Social Data Analytics at the Institute for Social Science Research at The University of Queensland, where she leads a second node of the Centre
for Social Data Analytics.

Livestream available

Lecture 3: Thursday 12 August 2021

Modelling to support the COVID-19 response in Aotearoa New Zealand

Dr Rachelle Binny, Manaaki Whenua - Landcare Research and Te Pūnaha Matatini

Mathematical models are playing an important role in the ongoing pandemic, providing insights into the spread of the virus and the effects of interventions to help inform response strategies. This seminar will give an overview of mathematical modelling by Te Pūnaha Matatini to support New Zealand’s COVID-19 response. We will describe the models used to simulate spread of COVID-19 in New Zealand, how they can help inform decisions on switching between Alert Levels, and how we are modelling the risk of new cases arriving at the border.

Rachelle Binny is a mathematical biology researcher at Manaaki Whenua - Landcare Research in Christchurch NZ, and a Principal Investigator in Te Pūnaha Matatini, the NZ Centre of Research Excellence for Complex Systems and Networks. Her research lies at the interface of mathematics, statistics and biology and is data-driven. Following a BSc in Mathematical Biology (University of Dundee, Scotland), she undertook a PhD (University of Canterbury, Christchurch) to develop new models of collective cell behaviour in wound healing, and calibrate these using experimental data. After completing her PhD in 2015, she spent two years as a postdoc at Manaaki Whenua (a Crown Research Institute for environment and biodiversity) before taking on a Researcher position there. Rachelle’s current research combines modelling theory with data from ecological systems to guide conservation management.

Livestream available

The triumph of the quants?: Model-based poll aggregation for election forecasting

Professor Simon Jackman, Chief Executive Officer at the United States Studies Centre, will examine recent successes and failures of predictive models of election outcomes. Professor Jackman will also discuss trends and discontinuities in the evolution of public opinion over election campaigns, spatial smoothing and pollster biases.

Watch the lecture

Machine learning for causal inference: Magic elixir or fool’s gold?

Professor Jennifer Hill from New York University will review the conceptual issues involved in understanding causal mechanisms and describe the potential for machine learning to improve our understanding of these mechanisms.

Watch the lecture

Implementing a machine learning tool to support high-stake decisions in child welfare: A case study in human centred AI (cancelled)

Professor Rhema Vaithianathan, from the Centre for Social Data Analytics at AUT, will reflect on what we can learn about applying data analytics in a trusted way, covering key concepts like consent, transparency, fairness and community voice, and how they can contribute to project success or failure.

Open source Machine Learning @ Waikato

Professor Bernhard Pfahringer from the Machine Learning research group at the University of Waikato discusses open-source Machine Learning software suites. He reflects on their design and their position in the current international Machine Learning landscape.

Watch the lecture

Deep learning: why is it deep, and what is it learning?

University of Auckland Professor Thomas Lumley discusses the rise of neural networks. He provides insight into how deep convolutional nets are structured and how they can be effective, but also why they are brittle and can fail in remarkably alien ways.

Watch the lecture

Algorithmic fairness: Examples from predictive models for criminal justice

Dr Kristian Lum from the Human Rights Data Analysis Group discusses the use of predictive models in the criminal justice system. Using examples from predictive policing and recidivism risk assessment she demonstrates how such models could perpetuate and potentially amplify data-encoded biases.

Watch the lecture

Statistical learning and sparsity

Professor Robert Tibshirani from Stanford University reviews the lasso method for high dimensional supervised learning and discusses some new developments in the area, including the Pliable Lasso, and post-selection inference for understanding the important features.

Watch the lecture

Myth-busting and apophenia in data visualisation: Is what you see really there?

Plots of data are important tools for observing patterns, but it is easy to imagine patterns that may not exist. Using two protocols the Rorschach and the lineup, Professor Dianne Cook of Monash University describes some simple tools for helping to decide if patterns are real.

Watch the lecture

Making colour accessible

University of Auckland Associate Professor Paul Murrell investigates the 'BrailleR' package for R and its difficulties with colour. By making a mountain out of that molehill, Paul embarks on a daring Statistical Graphics journey featuring colour spaces, high-performance computing, Te Reo, and XKCD.

Watch the lecture

Visual trumpery: How charts lie – and how they make us smarter

With facts and truth increasingly under assault, the use of graphs, charts, maps and infographics have become popular in supporting all manner of spin. Identifying information from misinformation is an important skill for any citizen. Alberto Cairo from the University of Miami teaches some guiding principles on how people can become more critical and better-informed readers of charts.

Watch the lecture

Expressing yourself with R

Hadley Wickham Chief Scientist at RStudio discusses Expressing yourself with R.

Watch the lecture

R and data journalism in New Zealand

Harkanwal Singh Data Editor from the New Zealand Herald on the use of R in New Zealand's data journalism landscape.

Watch the lecture

Interactive visualisation and fast computation of the solution path for convex clustering and biclustering

Genevera Allen, from Dobelman Family Junior Chair and Departments of Statistics and Electrical and Computer Engineering at Rice University, discusses clustering as a fundamental tool for exploratory analysis of big data.

Watch the lecture

Statistical computing in a (more) static environment

Ross Ihaka Associate Professor in the Department of Statistics at the University of Auckland discusses the spectrum of statistical computing systems from the dynamic to the very static.

Watch the lecture