Welcome to 2017 with 3 Data Science Talks

Join us back at out spiritual home for our first event of 2017.

There will be the usual beer 'n' pizza, networking and 3 x 25 minute talks.

Schedule:

6pm: Arrival. Please help us set up the chairs so we can start asap - then grab a drink (or 3) from the fridge and a slice or two of pizza.

6:20 - Welcome

6:30: - Using Data Science To Reach 10 Million Visitors Per Month (Marco Lui & Marton Bodonyi)

7:00 - 7:15 Break

7:15 - From student to 'scientist' - Starting out in data analytics (Joost van der Linden)

7:45 - Social Networks Analytics decoded and predicting deaths in Game of Thrones (Clement Fredembach)

8:15 - Questions & Networking

9:00 - close and head off to the Post Office Hotel ??

Talk 1 - Marco Lui & Marton Bodonyi

Using Data Science To Reach 10 Million Visitors Per Month

Melbourne-based transit search and booking site, Rome2rio, attracts up to 10 million visitors each month. In this talk, team members Marco Lui and Marton Bodonyi will share insights into using data science to drive your advertising strategy, to extract usable data on trends and patterns in your website traffic, to understand where you should be testing new UX functionality, and to delve into the complexities of A/B testing in a high-traffic consumer site.

Marco Lui completed a PhD in the Natural Language Processing group at the University of Melbourne in 2015, supervised by Tim Baldwin. He joined Rome2rio in October 2014, where he’s been working as a Data Scientist / Software Engineer. He is responsible for a diverse range of data-intensive analysis and engineering tasks.

Marton Bodonyi is a software engineer working in Melbourne at Rome2rio. He has over 6 years experience working in London and Melbourne on digital products and has recently started UsableTravel (https://usabletravel.com/), a blog that discusses UX in the context of the travel industry.

Talk 2 - Joost van der Linden

From student to 'scientist' - Starting out in data analytics

Working towards a PhD while simultaneously being employed as a "junior" data scientist has provided me with insights on what it's really like to progress from learning about data science to actually doing it yourself. You undoubtedly will have heard, or experienced, that learning and doing are not the same, but what is the actual difference in our sector? How do you go from writing assignments to communicating your data insights to your colleagues? What challenges will you face? In this talk, I will draw on my experiences of working with government and not-for-profit data to answer these questions and provide students and aspiring data scientists with a head start.

Joost van der Linden is a PhD student in Engineering at The University of Melbourne and a part-time data scientist at Our Community. Before settling Down Under, Joost obtained a BSc and MSc in Applied Mathematics from the Delft University of Technology in The Netherlands, and worked for IBM Research in America and for Schlumberger in England. Currently, he analyses grants and donation data for Our Community, a B-Corporation providing advice, connections, training and tech tools for not-for-profits and other organisations working to build stronger communities. Joost is also the founder and co-organiser of the yearly Melbourne Datathon.

Talk 3 - Clement Fredembach

Social Networks Analytics decoded and predicting deaths in Game of Thrones

This talk will use the book/television phenomenon to shed light into three problems often encountered in Advanced Analytics:

1. How to extract meaningful information from large text documents (e.g., contracts, books)
2. How to construct accurate social networks from text information
3. How to use Social Network Analytics (SNA) in conjunction with other propensity models to make accurate predictions

Using the 'Game of Thrones' book series, this session will introduce the analysis to:
• automatically parse and structure information to generate accurate social network of characters as the story develops; and
• make inferences on future character deaths using belief propagation across networks.

Combining the models allowed Clement to predict future death events with truly remarkable accuracy, demonstrating the predictive power of SNA alone and in conjunction with classical propensity models.

Clement Fredembach is a data scientist with Teradata Australia and New Zealand Advanced Analytics group. With a background in Colour Science, Computational Photography and Computer Vision, Clement has designed and built perceptual statistical experiments and models for the past 10 years.

Clement strives to combine his psychometric, perceptual and statistical knowledge to deliver insights and their story that are understandable and actionable to non-technical audiences.

Prior to joining Teradata, Clement collaborated with several Fortune 500 and academic institutions as a researcher, publishing and patenting large portions of of his research along the way.

Clement holds an MSc in Communication Systems from EPFL (Switzerland) on Image Classification and a PhD from UEA (UK) on Computational Imaging. His interests range from behavioral psychology to graph theory and photography.