COVER STORY

Facing Up To Data

Text by RICHARD HARTUNG & TAY QIAO WEI

Data has been heralded as the “new currency” of the digital age. How are our public agencies building up their capabilities in data analytics, and what more could be done to use data to improve citizens’ lives?
 
 cover-storycf85a97bf9f668da9107ff0000351172

Thanks to data analytics, new homeowners in Singa­pore can arrange a convenient time to collect the keys to their HDB flats through an online portal. 

Working together with the Government Technology Agency (GovTech), the Housing & Development Board’s (HDB) Estate Administration & Property Group ana­lysed approximately 100,000 emails they had received in 2015 about flat sales, to discover key topics of concern. They found that a significant number were requests to defer or reschedule the pre-specified key collection date. This finding led to an online booking system for all flat buyers to conveniently choose their preferred time slot for key collection.

This example is one of many ways in which public agencies can make sense of previously untapped sources of data to create meaningful change for citizens.

Making waves in many ways

Analytics has been rising in popularity and importance in recent years (read “Riding the big data wave” in Challenge July/August 2013). It has been a game changer in many fields including banking, human resources and fraud detec­tion, allowing us to identify patterns and predict outcomes.

Beyond analysing data in the form of text or numbers, agencies are now exploring video analytics to automatically derive insights from camera footage, instead of having hu­man operators spend hours scrutinising videos.

The Home Team, for instance, has been tapping video analytics to detect and respond to suspicious packages left in public spaces. “Video-trawling technologies can allow us to identify in a matter of minutes, objects and persons of in­terest from hours of video footage,” explained Mr Desmond Lee, Senior Minister of State for Home Affairs and National Development, in Parliament in 2016.

GovTech is also working with various agencies on a platform that will automate the analysis of video footage. This could help to track human traffic patterns or detect anomalies – enabling the Singapore Police Force to respond faster to potential threats, for instance. This video analytics system will be part of the Smart Nation Sensor Platform, which will enable agencies to collect, analyse and share data from sensors across the island.

Making sense of it all

To help the Public Service better exploit data, the Agen­cy for Science, Technology and Research (A*STAR) has gone beyond basic analytics to make sense of the in­sights gathered, says Professor Alfred Huan, Executive Director of the A*STAR Institute of High Performance Computing (IHPC).

Based on the data gathered, A*STAR scientists can build digital models to project possible scenarios, then use these models to develop simulations that can be used to study which interventions would work for each situation.

For example, by using commuters’ tap-in and tap-out information on train systems to simulate scenarios based on travel patterns and route choices, A*STAR worked with the Land Transport Authority (LTA) to predict and map out the distribution of commuters and their desired desti­nations. Such information will be helpful in the event of a breakdown, for example, for crowd prediction. “The crowds are not necessarily at the big interchanges like Bishan… Sometimes it’s counter-intuitive,” says Professor Huan.

These simulations have helped the authorities to pre­pare mitigation measures, such as providing alternative transportation or encouraging commuters to use different routes to relieve congestion, he adds.

Boosting the use of analytics

To achieve all that, and more in the future, the Public Ser­vice needs to grow capabilities to analyse data and develop simulations for insights.

Consulting firm McKinsey suggests in a paper Policy in the Data Age that governments need to attract talent into data-specific jobs such as data scientists, data archi­tects, user experience designers and analysts.

In the UK, the Government Digital Service is devel­oping a comprehensive talent-management strategy for digital and data skills. Not to be outpaced, the US has established DigitalGov University to train 10,000 federal civil servants every year.

Here in Singapore, Deputy Prime Minister Teo Chee Hean announced in Parliament in March 2017 that the government is building a core group of 250 professionals in GovTech and the Cyber Security Agency, who will have capabilities in areas such as data science and cyber security. Over the next four years, 10,000 public officers will be trained in digital capabilities so that they can use data sci­ence in areas ranging from policy formulation and service delivery to corporate services and cyber security. 

 

Democratising data science

Data scientists at both GovTech and A*STAR have created tools to make analytics more accessible for other public officers who are not trained in data science.

• GovTech has developed a text analysis platform that enables public officers to analyse textual data – such as emails from citizens – without needing to code. Known as “GovText (open to public officers only)”, the tool can help officers better understand what citizens care about.

• A*STAR scientists have created SentiMo, an analytics tool that analyses the tone of social media posts according to key words and phrases. The design concept and features of Sentimo have been shared with other agencies, as well as the private sector, to help organisations make sense of unstructured social media data in a faster and more reliable manner. 

bit.ly/sentiMo

emoji

 

Creating a data-driven culture

While it would be ideal to train everyone to code, “that’s impossible,” says Dr Daniel Lim Yew Mao, Data Scientist at GovTech. Agencies should instead focus on improving public officers’ data literacy, so that they can understand and be comfortable with how data science can be applied to their work, he adds.

Collaborations between GovTech’s Data Science Divi­sion and other agencies – such as the HDB – is one way to get there. Through these partnerships, Dr Lim and his team show officers how analytics can make an impact.

GovTech is also working on developing data science tools to make it easier for public officers, even those who do not know how to code, to apply analytics in their work. (See sidebox on page 10: Democratising data science)

In fact, GovTech data science specialists are placed in agencies such as the Ministry of Social and Family Develop­ment, the Ministry of Manpower and SPRING Singapore to build up analytics capabilities and culture in them.

SPRING Singapore Chief Information Officer Wong Ming Fai, for example, is seconded from GovTech. His small in-house analytics team is working to create greater awareness about analytics among staff and management, and where it can make a difference.

For example, they have created visuals to explain how the analytics engine makes recommendations. They have also started to brainstorm with different divisions on how analytics can be used to improve their work. SPRING Singapore also aims to send every staff for courses in ba­sic understanding of analytics, and some for more in-depth training.

For a start, SPRING Singapore has already begun to use analytics to better manage their programmes for SMEs, and track internal resource utilisation. Staff can also use analytics to identify which of the some 180,000 SMEs in Singapore they could work more closely with, based on fac­tors such as revenue and growth rate. This optimises funds and resources.

the-financial-impact-of-data

Translating data with behavioural science

Beyond having the technical expertise to gather and analyse massive amounts of data, what’s even more important is to be able to translate insights from data into meaningful ac­tions or change.

While analytics could be useful to guide certain services – say, to create new bus routes to meet public demand – or to predict what might happen in the future, it cannot change user behaviour. “The limitation is that data science doesn’t tell you how to solve the problem,” says GovTech’s Dr Lim.

For example, a hospital could analyse patient data to predict which patients are most likely to be readmitted – but this data alone will not prevent the phenomenon. Dr Lim shares that GovTech collaborated with Sing­Health to analyse electronic medical records, in order to identify patients at high risk of hospital readmission. The team created an algorithm that could predict, with 80% accuracy, the probability of each patient returning to the hospital.

The next step is to figure out what to do with the pre­dicted score. Dr Lim believes this is where behavioural science – which draws from psychology and economics to understand human behaviour – comes in. “We need to draw upon behavioural science to close that last mile in im­plementation,” he says.

The hospital could combine the predicted scores with targeted programmes to reduce the likelihood of readmis­sion. The question is, which intervention would be the most effective? Should the risk scores be shared with doc­tors to change the way they treat patients? Or should pa­tients be informed of their risk scores and be introduced to community activities that nudge them to lead healthier lifestyles? Trials to test out the various possible interventions could point the healthcare staff in the right direction, says Dr Lim.

More than data points

A*STAR’s Professor Huan also thinks that data science techniques should be “heavily complemented” with be­havioural science underpinnings. 

For those involved in creating, designing and commu­nicating social policies especially, a certain level of emotion­al intelligence and social awareness would be important. “As such policies are created for society, data needs to be contextualised at the human level,” he says.

For example, A*STAR’s IHPC scientists have used a social media sentiment analytics tool, SentiMo (see sidebox on page 10: Democratising data science), to analyse a par­ticular smartphone market and help a handset manufactur­er understand consumer sentiments on how its brand and products are received, what features are liked and needed, and how their competitors’ products are viewed. Similar technology could be employed to examine social policies and public perception of them, suggests Professor Huan.

And as our ability to gather and make sense of data continues to grow, it will be even more important to think of how we can make an impact on citizens’ lives, rather than get caught up with crunching numbers. 

Feeling the pulse of the economy

meter

Pulse of the Economy is a dashboard tool that allows agencies to use high-frequency big data – such as electricity consumption, online job listings or public transport data – to develop new indicators for economic and urban planning. For example, a dip in electricity consumption in a specific geographical area, together with a fall in the number of commuters at bus stops in the same region, could signal a drop in economic activity there. Public officers could then formulate strategies to counter this. The dashboard is an initiative by GovTech’s Data Science team, in collaboration with various government agencies.

 

bit.ly/GTpulse

Data with heart

How can organisations convince people to support worthy causes? Data could be a powerful tool. By customising email marketing to user profiles and behaviours, the National Volunteer & Philanthropy Centre (NVPC) increased public donation amounts for charities by 56% from April 2016 to April 2017. With a three-year visual analytics grant worth US$81,000 (S$112,000) from Tableau, a global visual analytics firm, the NVPC hopes to cultivate a “data-first environment” and further leverage data to promote philanthropy and volunteerism in Singapore.

bit.ly/nvpcdata

Share This Article