Matt Andrews
Software engineer making apps – that aren’t apps – and more at the FT. 会说汉语.

Big Data Analytics Tokyo 2017 — day 1

I recently attended Big Data Analytics Tokyo.

Here are my honest and unfiltered notes from the talks I attended on day 1:-

Building Innovation Ecosystems: What Can Tokyo Learn from Cambridge?

Overview - Slides

  • I liked Tim’s assertion that to be real innovation, an idea must be used by society, at large scale.
  • Innovation happens when you bring together Money Ideas and Talent (MIT).
  • Tim spent a long time emphasising the importance of co-location — people collaborate more with each other and the quality of the collaboration is higher when they work on the same floor. This was a slightly depressing message for me personally as I am working in a team that is split between London and Tokyo.
  • Boston (USA) seems like good place for all this data stuff.

Behind-the-Scenes Peek of an Analytics Startup

Overview - Slides

Honestly I really struggled with this. The topic wasn’t really relevant to my current projects and I think quite a lot was lost in translation — it was the first talk I’d watched with only simultaneously translation available via a headset (the previous talk being presented in English and Japanese) and slides were only in Japanese — I can read Mandarin Chinese reasonably well, so I could understand some of the Japanese Kanji so I think I had an advantage over other non-Japanese speakers but it was still pretty hard.

I think the core messages were:-

  • Analytics firms would be best to target their products at the marketing department first — to help them achieve business results by optimising something.
  • In Japanese firms people performing data or technology leadership roles are substantially less likely to report to the firm’s CEO. Is the implication that analytics has less importance/a lower status?

The New Vanguard for Business Connectivity, Design & The Internet of Things

Overview - Slides

Probably the most enjoyable of all the two days of talks.

  • I loved the term ‘enchanted objects’ as a category of ‘Internet of Things’.
  • Image recognition is really powerful now. We can literally buy the clothes our friends have been photographed wearings, or buy flights to the locations our friends have taken photos at and more, …
  • Will we have sarcastic dustbins that watch what we throw away, reorders our groceries or warns us for eating too many cookies?

Uncovering Team Performance Dynamics with Data & Analytics in Complex Engineering Projects

Overview - Slides

This was introduction some academic research into optimising large scale projects. For example, if project’s members are split across multiple regions with multiple time zones what would be the most effectively way to split up tasks between the various locations. The assertion being a project’s costs could be dramatically reduced with the clever utilisation of software:-

We will be able to predict and provide teams with real-time adaptive tools and thinking leading to great performance.

My personal experience of large creative software engineering projects, which can be very unpredictable and even assigning the same task to different engineers within the same team can lead to quite difficult results, so I felt somewhat sceptical about the conclusions on offer here.

The Dirty Little Secret of Enterprise Data

Overview - Slides

Although the talk was a little bit ‘a word from our sponsors’ there were some useful ideas I took away from it…

  • The secret is that data is silo’d into different systems and owned by different teams.
  • Cleansing and organising data is the biggest challenge for companies, which splits into 3 systems:- matching records, classifying items and mapping columns/attributes.
  • You could use machine learning to merge data sets together.

Big Data Analysis for Cyber Security

Overview - Slides

Key takeaways:

  • Cyber Security is a great use case for Big data and Machine Learning.
  • Recommendation algorithms can be used to detect infected devices (and similar devices).

The Investor’s View of Emerging Data Marketspace in Japan & the US

Overview - Slides

Some common sense suggestions for aspiring data entrepreneurs in Japan:-

  • focus on solutions that bring data ubiquity — making data accessible, shareable, social …
  • learn to dance with elephants — work with big companies, recognise data is very valuable, build trust, …
  • ‘multi-sided markets’ — I think the point here is to deliver value to multiple stakeholders at once (helping the client and helping the customers)

Things to worry about:-

  • Don’t miss the boat
  • But don’t leap before you look
  • And don’t get squashed (by the elephants that you’re dancing with…)

Artifical Intelligence Sparks the Fourth Industrial Revolution

Overview - Slides

Another ‘a word from our sponsors’ talk but I didn’t really take anything away from it.

Data Science Initiatives at a FinTech Company

Overview - Slides

I really struggled with the language issues here again — the slides were almost all in Japanese. That said, it was fascinating to learn about a highly successful Japanese FinTech startup.

The speaker’s favourite algorithms were Support Vector Machine and State Space Model.

And finally.

  • The venue was breathtaking. Taking place on the 49th floor at the Roppongi Hills Academy, we were greeted by stunning views of Mount Fuji in the morning and glorious sunsets and nightscapes in the evenings.
  • If I ever do a conference talk here I should make sure the slides are bilingual.
  • To get the most out of my time in Japan I really need to learn more Japanese.
  • Many of the western speakers presented in Japanese. I would be curious to learn whether it was better from the Japanese-speaking audience’s perspective to hear talks presented in Japanese as a 2nd language or in native English and simultaneous translation.
  • I am inspired and want to learn more about AI, ML, NLP and Big Data.