EVENT - Tuesday
LocationDock House, The Quays, Salford Quays, Salford M50 2LH
When6:00 PM - 9:00 PM
DSF Meetup with the BBC
Join the Data Science Festival Manchester in partnership with the BBC for 4 epic talks this July. From Predicting School Holidays Using Video Streaming Data to adapting to cope with our data science teams 300% growth, all the way to Learning from archives: how historical content can be used to engineer new content production tools.
Due to the popularity of Data Science Festival events, we are now allocating event tickets via a random ballot. Registering here enters you into the ticket ballot for the Data Science Festival Event on July 16th 2019, the ballot will be drawn on the 12th July 2019. Those randomly selected will then be e-mailed a Universe ticket for the event.
If you get an allocated Universe ticket, please bring a copy of your paper ticket or your ticket on your phone to the event to check in with your QR code. Tickets are non-transferable.
PLEASE NOTE REGISTERING ON MEETUP DOES NOT GUARANTEE YOU ENTRY TO THIS EVENT.
Please click here to apply for a ticket: GET TICKETS
6:00pm: Doors open
6:30pm: Dr Matt Crooks
6:50pm: Heather Miller
7:45pm: Adam Dicken
8:10pm: Tamsin Nooney
Dr Matt Crooks – Predicting School Holidays Using Video Streaming Data
Summary: Matt will show how we can leverage user behaviour through various clustering and classification techniques in almost real time to deliver better recommendations to our users.
Bio: Matt Crooks is a Data Scientist in the Central Insights team. He has a PhD in mathematics in which he attempted to model and predict earthquakes. After completing this PhD, he took up a postdoctoral research position at the University of Manchester as a Climate Scientist for the European Commission. His work involved using data collected from laboratory experiments and field campaigns to develop mathematical models to improve the representation of clouds in weather and climate models. During this time he worked closely with the Met Office and has seen some of his work encapsulated into their forecasting models, although he accepts no responsibility for incorrect weather forecasts! Upon leaving academia at the end of 2017, Matt joined the BBC as a Data Scientist, where he has been helping product teams in CBBC and Bitesize to understand their audience and improve the user journey on the websites.
Heather Miller – Experimentation in the BBC: Statistics For The Internet Age
Summary: This talk outlines how we test at scale across the BBC, including the tools we utilise along with the type of content we test on a daily basis. The final part of this talk outlines some of the statistical challenges with testing at scale in real time, and some of the solutions we use.
Bio: Senior Experimentation Analyst
Adam Dickens – Best Practice as Code: How we are adapting to cope with our data science teams 300% growth
Summary: Adam will present several ways we are utilising best practices from software engineering in our data science team, including;
– Encouraging best practice with templates (packaging using Docker / virtualenv, testing, project structure, etc)
– DRY data science – Making commonly used routines available as a library
– Making automation easy
Bio: Adam got his masters in Physics, going on to build recommendation systems at Totaljobs before coming to the BBC in 2016. Adam loves to travel and recently spent a year travelling around the world with his fiancé, eventually finding his calling milking cows in New Zealand, but ultimately returning to the BBC (because cows are too much hard work…). Adam’s main focus is technology within the data science team.
Tamsin Nooney – Learning from archives: how historical content can be used to engineer new content production tools.
Summary: There are now many tools to enhance video archives with extended metadata such as shot boundaries / types, face locations / identities etc. Here we will present examples of how large quantities of historical metadata can be used to develop new tools that would have been prohibitively difficult previously. In particular we will present work on Ed a system which we have developed to automate the coverage of live events with a multi-camera setup. We will show several examples of how the system has learnt how to frame, choose shot type and time changes by examining features in historical video such as face location, shot type, shot timing and audio features