Speakers at the Analytics Summit 2019


Meet the Speakers

Steve.png

AS19 Plenary Speaker: Steve Caughey

The OR Society is delighted to announce that Steve Caughey, the Director of the National Innovation Centre for Data, will be joining us as Plenary Speaker on the topic of the centre and its work.

This £30M investment brings together industry, the public sector and world-leading academics to develop the skills, ideas and resources needed to exploit the opportunities offered by the explosion in digital data. In short, the centre helps organisations get the skills they need to obtain insight from their data. In his talk, Steve will further explain the purpose of the centre and how it intends to work with public sector organisations to build a shared data ecosystem, help upskill employees, share best practice and initiate cross-organisational data projects.

Steve worked as a software engineer and systems architect for a number of telecommunications companies before joining the influential Arjuna research team at Newcastle University in 1990. He was instrumental in commercializing the prototype work of this group by co-founding Arjuna Solutions in 1998. Steve led this startup company to acquisition, by Hewlett-Packard in 2000, bought back the company (renamed as Arjuna Technologies) in 2002 and sold part of the product suite to RedHat in 2005. As CEO he was responsible for strategy, corporate development and partnerships.

Payal.png

Extracting Value from Data Analytics: Focus on People, not Data

Payal Jain, MD JCURV & Chair Women in Data, UK

Abstract: SAP found that 86% of organisations believe there is further value to be extracted from their data; but 74% of the organisations surveyed believe it’s too complex to monetise this value. It makes for bleak reading for the boardrooms that were persuaded to make these investments. So how can Chief Data Officers (CDOs) turn this ship around?

This talk will define the building blocks for an effective data analytics function. Real life case studies from FTSE 100 firms will be shared where they have transformed their data function by focusing on people, not data. With the advances in AI and the focus on bias and ethics, this talk addresses that the skills sets required in data moving forwards and what are our roles as data leaders will be in the future.

Bio: Payal is a Managing Director at JCURV, a management consultancy company who’s mission is to increase the agility of UK PLC by helping companies radically change the way they leverage data.  Previous to this, Payal has held several executive roles in the Banking and Finance industry for the last 18 years including Existing Customer Management and Loans Director at Vanquis Bank and Managing Director of Strategic Analytics at Barclays, responsible for all the analytics in the UK and European Credit Card and Lending division across 8 countries. 

Payal is the Chair of Women in Data, who’s mission is gender parity in all data roles, especially at a senior level.  In 2016, Payal was recognised as the most influential data professional in the DataIQ Top 100 leaders in the UK and shortlisted as Woman of the Year in the 2019 Credit Awards.

Erik.png (1)

Making AI Human Again: The Importance of Explainable AI (XAI)

Philip Pilgerstorfer, Data Scientist, QuantumBlack, a McKinsey Company

Abstract: Proliferation of machine learning techniques have led to more companies driving their decision-making process with complex models. In this increasingly fast paced world, it is becoming more difficult to understand how AI systems are making our everyday decisions. Along with increased regulation, there is a growing demand to build explainable models without compromising performance. In this session we will present how QuantumBlack is using XAI to drive adoption, diagnose bias, and deliver impact across organisations.

Bio: Philip Pilgerstorfer is a Data Scientist at QuantumBlack. He has delivered projects in manufacturing, oil and gas, motorsports and pharmaceutical. He is a contributor to QuantumBlack's internal R&D efforts like Explainable AI (XAI) and on causal inference. Philip's academic background is in econometrics, statistics, and machine learning.

Didier.png (1)

Ethics and Fairness in AI: A Systemic Approach

Giles Colclough, Data Scientist, QuantumBlack, a McKinsey Company

Abstract: This workshop will be focused on ethics and fairness, areas that are becoming increasingly more important for industrial applications. There will be knowledge sharing between industry and academia through a use case, which will highlight the great opportunities between quantitative accuracy, ethics and fairness and how to manage this potential trade-off. The session will enumerate the critical underlying risks, mitigations and practical checklists that have been formalised at QuantumBlack.

Bio:

Giles is a Data Scientist at QuantumBlack, where he works with clients in the healthcare and pharmaceutical industries to realise the value of their large data sets. Giles believes that the quality and cost of patient care can be radically transformed by capitalising on the wealth of data available in these sectors. However, it is vital - particularly where algorithms will inform decisions taken about patients or their treatment - that any prediction or recommendation systems that are deployed are both interpretable and unbiased. He therefore joined a team at QuantumBlack that has developed tools and processes to support the deployment of fair and transparent machine learning models. 

Giles holds a DPhil from the Oxford Centre for Human Brain Activity, where his research contributed new methods for modelling connectivity in the human brain. He has a degree in physics from the University of Cambridge.

Cosh.png (1)

User-centred Analysis and Visualisation

Emma Cosh, Analytics & Data Visualisation Consultant

Abstract: If the purpose of analysis and visualisation is to inform people and drive actions, how do you ensure your work has the best chance of succeeding. This talk introduces a framework and some tips and tricks for putting users at the heart of data visualisation and making it as engaging as possible.

Bio: Emma is a freelance analytics and data, visualisation consultant. Her work has been featured in The Guardian and Tor.com, and been longlisted for The Information is Beautiful Awards. She started out in operational research and statistics, and gradually shifted from building models and leading analytics teams to communicating and visualising data, making it easy to understand and beautiful to look at.

Herbert.png

Movement Insights – Measuring the Performance of Bath’s High Street

Allison Herbert, the CEO of Bath Business Improvement District

Abstract: Bath Business Improvement District (BID), together with city stakeholders and Bath & North East Somerset Council, are pioneering the use of Movement Insights, a new SaaS product developed by Movement Strategies Ltd. The web-based dashboard provides the BID access to a range of tailored insights to measure the performance of the city centre from a new perspective, focusing on the behaviour of people including residents, employees and visitors.

Movement Insights brings together a number of rich data sets including Mobile Network Operator (MNO) data, credit/debit card transaction data, footfall data, public Wi-Fi data and sentiment analysis. The paper provides a brief overview of the data sets, the methodology and the product, as well as focusing on the value of the insights to the Bath BID and their stakeholders.

Bio: Allison Herbert has a background in economic regeneration, working across deep rural communities, market towns and the city centre of Bath from within and outside of local government.  As Chief Exec of Bath Business Improvement District (BID), she is responsible for a programme of activity and campaigns which deliver tangible benefits to 700 city based businesses.

Babes.png

Movement Insights – Measuring the Performance of Bath’s High Street

Simon Babes, Managing director at Movement strategies

Abstract: Bath Business Improvement District (BID), together with city stakeholders and Bath & North East Somerset Council, are pioneering the use of Movement Insights, a new SaaS product developed by Movement Strategies Ltd. The web-based dashboard provides the BID access to a range of tailored insights to measure the performance of the city centre from a new perspective, focusing on the behaviour of people including residents, employees and visitors.

Movement Insights brings together a number of rich data sets including Mobile Network Operator (MNO) data, credit/debit card transaction data, footfall data, public Wi-Fi data and sentiment analysis. The paper provides a brief overview of the data sets, the methodology and the product, as well as focusing on the value of the insights to the Bath BID and their stakeholders.

Bio: Simon has delivered a significant number of high profile projects over the last 20 years in the UK, Asia and the Middle East in the transport, sports & entertainment, cultural and smart cities sectors. Simon has led the development of Movement Insights, a new SaaS product delivering insights derived from multiple 3rd party data sets. Prior to joining Movement Strategies, Simon established successful consultancy businesses in China and Malaysia having studied Engineering at Durham University and started his career with the Ministry of Defence and then moving to London Underground Ltd. He is currently a Board member of the UK’s Railway Industry Association (RIA).

Katarzyna.png (2)

Workshop: Tutorial on explainable machine learning in R

Kasia Kulma, Senior Data Scientist at Mango Solutions

Overview: Machine Learning has seen an unprecedented development over the last decade and offers a great promise of solving various predictive problems. However, the most accurate models are often black-box, i.e. they don’t provide an explanation why they made a particular decision. This uncertainty is undesirable not only from the technical (model improvements) or legal point of view (e.g. GDPR). It can also introduce bias in high-stakes decision that operate at scale (loan & credit card applications, promotions and recruitment, sentence parole). In order to gain/maintain trust in model predictions it’s crucial to have access to model’s interpretable explanations.

Prerequisites: This course is designed for practitioners who want to get a better understanding of their ML models and get better knowledge of popular explainable ML frameworks. Attendees should:

  • write R code at intermediate level
  • have a basic knowledge of binary classification models
  • familiarity with ML model training, testing and evaluation workflow

Outline: In this tutorial we are going to show the pitfalls of trusting models based on their accuracy alone. Then, we will introduce LIME (Local Interpretable Model-agnostic Interpretations): a model-agnostic frameworks for explainable ML. We will show how to apply this method to standard classification problem and how this may change how much we trust original model predictions.

After the course, you will be able to:

  • Approach with caution pure accuracy measures.
  • Understand local drivers behind correctly- and –incorrectly predicted instances (LIME).

Bio: Kasia Kulma is a Senior Data Scientist at Mango Solutions and holds a PhD in evolutionary biology from Uppsala University, Sweden. She has experience in insurance, delivery and environmental services industries. She is the author of the blog r-tastic.co.uk and is a mentor and organiser in R-Ladies London. She is an R-enthusiast interested in data (science) ethics and evidence-based medicine.

Hannah.png

Workshop: Tutorial on explainable machine learning in R

Hannah Frick, Data Scientist at Mango Solutions

Overview: Machine Learning has seen an unprecedented development over the last decade and offers a great promise of solving various predictive problems. However, the most accurate models are often black-box, i.e. they don’t provide an explanation why they made a particular decision. This uncertainty is undesirable not only from the technical (model improvements) or legal point of view (e.g. GDPR). It can also introduce bias in high-stakes decision that operate at scale (loan & credit card applications, promotions and recruitment, sentence parole). In order to gain/maintain trust in model predictions it’s crucial to have access to model’s interpretable explanations.

Prerequisites: This course is designed for practitioners who want to get a better understanding of their ML models and get better knowledge of popular explainable ML frameworks. Attendees should:

  • write R code at intermediate level
  • have a basic knowledge of binary classification models
  • familiarity with ML model training, testing and evaluation workflow

Outline: In this tutorial we are going to show the pitfalls of trusting models based on their accuracy alone. Then, we will introduce LIME (Local Interpretable Model-agnostic Interpretations): a model-agnostic frameworks for explainable ML. We will show how to apply this method to standard classification problem and how this may change how much we trust original model predictions.

After the course, you will be able to:

  • Approach with caution pure accuracy measures.
  • Understand local drivers behind correctly- and –incorrectly predicted instances (LIME).

Bio: Hannah Frick is a Data Scientist at Mango Solutions and holds a PhD in statistics from Universitaet Innsbruck, Austria. She has authored and maintains several R packages on CRAN. She's a co-founder of the R-Ladies Global organisation and part of the leadership team. You can follow her on Twitter @hfcfrick.

Emma.png

Introduction to Pricing Analytics

Emma Murray, Senior Managing Consultant in IBM Services

Abstract: For most organisations, pricing is the most powerful lever to drive revenue and sales. In this introduction to pricing, we will explore how systems thinking can be used to define the right KPI’s to track pricing success.  We will also examine how data science can help organisations to better predict the impact of price changes on revenue and how OR techniques can identify the best price to deliver improved profitability.

Bio: Emma Murray is a management consultant in IBM’s Services division, specialising in the use of advanced analytics to solve industry problems for retail and CPG clients. She has worked with a number of large organisations across Europe, with a key focus on pricing and revenue management. Emma is also a certified pricing professional in the Professional Pricing Society.

Eduardo (3).png

Agile Analytics: Using R Shiny to develop data analytics tools 

Eduardo Contreras Cortes, Data Analytics Manager at EY

Overview: Agile software development is an approach to software development under which requirements and solutions evolve through the collaborative effort of self-organizing and cross-functional teams and their customer(s)/end user(s). It advocates adaptive planning, evolutionary development, empirical knowledge, and continual improvement, and it encourages rapid and flexible response to change. 

Borrowing this approach and applying into the Analytics context, the workshop objective is to  develop an analytics tool using  R and Shiny as an user interface that will load data, analyse descriptive statistics and interactive plots to then calculate machine learning models and let the users download the results all under the same environment

Bio: Eduardo Contreras Cortes is an actuary and statistician with more than 10 years of experience in consulting and financial services. Currently he works as Data Analytics Manager at EY and he is the Chair of the Analytics Network, part of the Operational Research Society.