At the Strata Conference in London this year (2015) the topics of Big Data, Machine Learning and Analytics were all exploding as technologies demanding attention. It feels a lot like an industry finding its way, but the level of energy will mature the thinking pretty fast. I had seven key take-aways and points of interest.
Infrastructure and Business Strategy
Business Strategy and Change Management was under-represented. There were just a couple of sessions – from Silicon Valley Data Science, and from Accenture. Exhibitors did not show clear user cases, nor how individual choices map to business strategy. At times it felt as though heads of IT were there to find solutions to unknown problems. It seemed as though strategy in this area is a new practice. Three sessions discussed how to think about business and data strategy. They emphasised the importance of a prototyping team to bring new ideas, methods, and change into an enterprise – to accelerate and bypass the difficulties of an incumbent talent mix and political silos.
It ain’t what you do to data, it’s what you do with it (PPT)
The data strategy revolution: building an in-house data insights lab
Understanding the Chief Data Officer
Machine Learning and APIs
A growth area is Machine Learning process pipelines. Following their buyout, AlchemyAPI are now integrated as a Watson service into IBMs product BlueMix. Developer tools are set up as microservices which can be added together into a pipeline, and which grow into a sophisticated API. Payment plans are per microservice and per microprocessor RAM/Cores. This kind of solution lends itself well to exploration and agile iterative product design. I had valuable conversations with lead data scientists from IBM Watson and Dato. Data also put together a simple and pleasing recommendations engine for Session similarity, on the link below.
IBM BlueMix console
Dato’s session recommender
Dato’s learning gallery
Graphs and ecosystems
This wasn’t represented particularly well as ‘a thing’, although much of the time it is an assumed technology. The best insight for me was from AutoDesk, who have built a ‘Design Graph’ to interconnect parts (nuts and bolts) and functional systems (engines) into final objects (cars). The Relationships in the graph show the combinations by which sets of things become Systems. The talk is below, and the design graph slides are at the 3/4 mark.
AutoDesk’s Design Graph
Old and new infrastructures
I discussed new and old technologies, and the challenges of building new things with an Oracle representative. He noted that a shortage of developers has led many companies to use a pragmatic old+new hybrid model. I suggested that a Graph should hold a ‘model’ of the whole system and act as a reference meta-store. That way you could link legacy systems and SQL-based systems (where the stuff lives) with analytics loops to augment the graph.
Data for Measurement
There was a good session by ScraperWiki’s CEO who discussed the UN’s Human Development Exchange (HDX). Designed and built by Frog, this site uses technologies from the Open Knowledge Foundation (CKAN) and ScraperWiki. There’s inspiration here for ‘Measurement’ and ‘Solutions’ work.
The Ebola issue on HDX
From the perspective of presenting ideas to an audience, two things were clear. The talks which tried to tell ‘a story’ fell flat. The TED format, if not done well, can seem tired and manipulative and lacking real advice. But talks that used slides showing a process, or a way of thinking, or an implementation plan, or a diagram that shows how concepts come together – brought out camera phones every time.
People are hungry for ‘thinking models’ and strategy. A new format for Keynotes with Case studies starts to address a strategy-implementation gap. The head of IT at BT discussed the how and why of building a multi-tenant hadoop platform in response to real speed and legacy system
How BT Uses Hadoop with its CRM
Data, Privacy and UX
There was an outstanding presentation from Sentiance’s Ann Wuyts on the link between Privacy and UX Designing for Opt-in, No Surprises, Consent, and Trust.
View Ann’s presentation on Slideshare