NLA Ltd was delighted to be ask to join a panel on Big Data at the recent Digital Ship CIO conference in Singapore.
This is a hot topic, and will only increase in importance when you consider that the data analysis market as a whole is predicted to grow to $203bn by 2020, up from $130bn in 2016.
While acknowledging that there is great potential for Big Data within shipping, we were keen to point to other sectors for examples of where data had been used successfully. We present below a small selection of the dozen case studies we referenced on the day.
1. Size doesn’t (necessarily) matter
An important starting point is that this debate should not solely concern itself with ‘Big’ Data. Business Intelligence and CIOs have always used data, and while the volume, variety, velocity and veracity of available data has grown exponentially in recent years, that can be as off-putting as empowering. What matters is the insight gained and the potential actions suggested, not the size of the data set.
An example of the power of small data can be found in the education sector. A couple of years ago, a Washington DC elementary school conducted a short study. The school had been using tablets as an aid to teaching and learning, so invited in 24 volunteer data scientists to see if they could find anything interesting in the system’s data.
In just 24 hours, they highlighted which groups of students benefited the most from such programmes. On a reading app, identified patterns showed what type of usage was most likely to lead to increases in literacy scores, differentiated across age groups. On a typing game, they discovered that students either aimed for speed or accuracy, with the latter showing a 36% improvement in typed words per minute, more than double the 17% recorded improvement of the “fly-through” students. These and other insights provided teachers with greater understanding of usage, and suggested actions to improve learning.
Quite rightly, analysts talk about how massive data sets can offer transformative gains for businesses. However, advances in data analysis techniques can also be usefully applied to much smaller issues, with much smaller data and at much less cost.
2. It’s not just numbers
The chair of the session (the excellent Saurish Nandi, Founder and CEO of Proternio Consulting) challenged the panel to suggest why – if so much data is now available – so many collisions still occur at sea?
Two things struck me here. First of all, in conducting my own quick research, most conversations concerning this seem to be happening in academia. Could some of the answers be stuck in a paper somewhere? Piotr Borkowski’s paper might be a good starting point to explore the issue.
More broadly, though (aside from the emerging interest in deploying autonomous systems in this area), case studies in other domains repeatedly highlight that the answers to problems may lie in forgotten or uncorrelated data sets which means that organisations need to understand what data may or may not be useful.
Perhaps the most celebrated case is where New York City authorities used predictive modelling in 2011 to pinpoint buildings with a higher than normal fire risk. Whereas previously they would only have looked at standard data such as building age and historical fire risk, the new model attempted to pull in a much broader wealth of base information. They focused on factors like missed tax or utility payments – which suggests neglect – and nearby crime and accident rates. Then, by visiting only those buildings that the data pointed to, fire inspectors were able to find many more than if they'd simply gone door to door. The model has continued to progress in the subsequent years.
So, when looking to data science to solve long-standing problems in any industry, one of the early questions should also be: what data have you got, even if at first glance it may not seem to be relevant?
3. Data leadership is essential
Someone has to lead.
Deriving useful insights from data – in any industry – is difficult, will take time, and may provide lots of false promise before delivering value. However, talking the talk without supporting CIOS and other related managers with appropriate resources will only lead to failure. Data leadership – coming from the very top levels of management – will help to provide proper foundations for success.
We can look across to the art world for an example of some head-turning data leadership. In 2014, the Dallas Museum of Modern Art took an unprecedented step. They made admissions and memberships completely free, as long as patrons were willing to share some personal data – even just their name and email address.
This was an initiative with a significant price tag – the museum was immediately wiping out $1m from its bottom line, which equated to 5% of its annual budget. However, when announcing the initiative, Director Maxwell Anderson said, “We’re trying to incentivise people to represent what they’re doing, where they’re going, and how they’re spending their time.” This was, clearly, a fundamental and very public change in approach to the value of data to the organisation.
The result? Membership tripled in a year. Through associated initiatives, museum staff were also able to track which galleries were most popular, which members were repeat visitors, and what low income areas were being served (or not served) by the museum most. It was immediately clear to management that this type of data is absolutely critical for grants and fundraising.
As for the finances, free access to the museum has increased visitors – who spend money in the café, gift shop, and on special paid programmes. In addition, the museum has been able to use the 2 MILLION data records to raise $5 million to support the museum (which is, not to forget, 5 times what the revenue was for paid admission and memberships).
Not every business will be able to undertake such leaps of data faith, but none will be able to do so without the full support of senior management.
The shipping sector is on its own data journey. With more case studies of the profitable use of Big Data within shipping emerging, leading to greater understanding of future potential, and with more calls for greater data sharing in the supply chain, the next few years will hopefully see great progress.
We thank everyone at Digital Ship for inviting us to participate in the Singapore conference.