How data science is helping charities to fight hunger in the UK

This article - reposted from Guardian Voluntary Sector Network - highlights an AAM project that allows The Trussell Trust to map current demand, highlight places with unmet need and predict future patterns.

The Trussell Trust’s latest statistics show that food bank use in the UK remains at record levels, and that the supply of emergency food parcels has increased by 2% in the last year.

Amid the passionate discussion that these figures arouse, one comment by David McAuley, the trust’s chief executive, deserves attention: “Reducing UK hunger will require a collective effort from the voluntary sector, government, the Department for Work and Pensions, businesses and the public. The Trussell Trust is keen to work with all these groups to find solutions that stop so many people needing food banks in future.”

An admirable aspiration, but how can you unite these disparate agencies around a problem with such jagged edges and complex causes and effects? I believe the answer might lie in data science.

Over the past 18 months, the trust has been working with researchers from Hull University Business School and us at social innovation agency AAM Associates to explore how new technologies can help them to fight UK hunger.

Using technology

The initial research phase identified 56 technologies that are already being used in food banks globally, and we held workshops with Trussell Trust staff to identify what might help them most with their operational needs. While digital fundraising methods and volunteer management systems were of great interest, most ideas felt like “nice-to-haves” rather than game changers. There was one exception – data analysis.

Charities all over the world, such as Citizens Advice and the South Carolina Campaign to Prevent Teen Pregnancy, and public bodies including the New Orleans Fire Department, have shared encouraging results from data analysis projects.

This aligned well with the trust’s More Than Food initiative, which looks beyond providing emergency food and towards tackling the underlying causes of hunger and poverty. To do that, they need to know more about the people who come to them for help.

The trust’s core data includes food bank locations and individual client data, such as their names, addresses, ages and underlying causes of crisis – benefit delays, school holidays, homelessness, etc.

With data science firm Coppelia recruited to the project, and having taken advice directly from the Information Commissioner’s Office on appropriate data security, our initial analysis highlighted some noticeable regional variations and delivery patterns year on year.

On the map

A mapping tool was used to show how the trust’s users are distributed geographically. Heat maps showing demand – the darker the colour, the greater the demand – have allowed us to visualise regional patterns of need.

Even a cursory glance at the map suggests countless questions: how far do most clients travel? Are food banks located in the areas of greatest need? Why are particular crisis types more prevalent in certain areas?

More creatively, the platform also looks to predict potential areas of “unmet” need. The data was crunched to show the number of people fed by the trust, per head of population, in each local area since 2014. Various census statistics (levels of deprivation, unemployment) were then used to pinpoint key characteristics of the areas that received the most support. The results were used to predict – again using heat maps – where need for the trust’s services may be greatest across the country.

The model is not perfect, but it already poses the question: how might the charity respond to signs of unmet need?

What next?

For charities – especially those on the frontline helping people in crisis – this project highlights the importance of collecting data, and how the potential for aligning open data sets (a charge led well by the Open Data Institute) with that data is rapidly becoming a reality.

Data scientists can be expensive, but data-hungry academics are always looking for frontline partnerships, and bodies such as the OR Society run great pro-bono programmes to help.

There are two next steps for the Trussell Trust in relation to this work – and neither is easy. First, they need to answer some of the many questions highlighted here. Data can provide new insights, but how do you act on them? How can you change focus, practice or even business model to respond to the issues?

More strategically, the trust is beginning to look outside of its own walls for solutions. Sharing data with other charities – safely, securely and with the appropriate permissions – could create a significant step forward in understanding and answering local need.

The causes of poverty are complex, brittle and often unique to the individual case. Each cry for help, each visit to a food bank, each call to a debt advice line, paints an individual brush stroke in a much richer picture of local need.

Pooling data might provide users of crisis services with some small comfort that they are helping to deepen this country’s understanding of poverty, and in doing so, perhaps shielding others from its harsh glare.

Research Workshop Report - What's Digital about Fashion Design?

FIRE and AAM collaborated to explore how technology is used within the designer fashion community. We wanted to identify how traditional fashion design is being transformed by digital technology, and to see how willing designers are to adopt new digital methods and models.

We invited established fashion and textile designers, academics and industry experts to take part in a workshop, and asked them to interrogate the question ‘What’s Digital about Fashion Design?’. We wanted the workshop to be a space where industry professionals could discuss how the fashion industry currently operates within the digital economy, and discover where opportunities for new models in both product and business development could come from. Here’s what we found.

“The fashion industry likes how it works, it has an entrenched way of operating – the seasonality is very difficult to change.”

The first theme that emerged was The Challenge to Tradition. We deconstructed the traditional fashion design cycle and questioned how the transition from wholesale to direct consumer retail is changing fashion design. It was agreed that whilst digital activity is mainly focused on sales, there is an unexplored opportunity to introduce digital in the earlier stages of the design process, which could have the potential to transform both a designer’s practice and business. We acknowledged that while it is important to challenge fashion traditions, we didn’t want see digital entirely replace craft methodologies, but to be integrated in a way that would enhance a designer's practice and raison d'être.

“Digital technology obviously isn’t going to be a replacement for craft-based design practices, it should be more about how it can enhance what I want to achieve.”

Our second theme outcome, Digital and the Design Process, challenged the potential of digital interventions to help save time and costs in day-to-day processes. It is a well known fact that fashion designers are time-poor, fast-paced, and on a tight budget, and these concerns were strongly shared at our workshop. Conversations explored how designers could potentially invest more time and money into creative research and development processes if physical overhead costs were cut and replaced with digital services.

“What does it mean to embed digital into operations for the designer fashion community? Especially when you no longer need a physical store?”

The idea of removing the physical store was just one proposed solution, and by all means not for everyone. However, by removing the high cost of company stores, designers would be able to embark on new journeys and perhaps bring in new team members such as consultants and developers. Replacing the physical with digital would greatly affect budgets and free up spending for embedding digital into day-to-day operations. The thought of this seemed fairly plausible and exciting!

“We need to collaborate, open research that is the only way things will change and designers will start to define their place in the digital economy.”

We discussed how collaborations between fashion, technology and manufacturing are needed to successfully integrate technology into design and process, and how the potential to house designers and tech start-ups in the same building might organically spark these types of collaborations and knowledge sharing. We speculated about a more ‘start-up’ culture in fashion, borrowing working practices from the entrepreneurial tech industry who, unlike fashion designers, are not wedded to the traditional fashion design cycle and are seen to be driving the development of new revenue models.

Developing new models was our final theme. By the end of the workshop we felt that fashion designers were considered to be in a unique position, with specialist knowledge and understanding of design to help shape developments in digital technology. There was a shared cautiousness about adopting digital business operations as there is a lack of prevalent technology in fashion supporting the transformation. We are still left with the question as to what extent digital engagement could be integrated into the entire product development lifecycle within the present fast-paced business and slow cash flow models.

“There’s a lot of ‘We know digital is important, but…’- Not many designers are able to invest time, energy and money in understanding what digital services would work for them. They’re too busy getting ready for the next season!”

Our workshop felt like just the beginning of questioning, understanding and breaking down the barriers and exploring opportunities for new models of practice with digital technologies during this period of significant growth in the digital economy. Digital will continue to play an increasingly important part of our lives and the workshop showed how willing the SME designer community is to use technology to serve both the design process and product and business model development.

Below is a link to our report, which goes into our research in a lot more detail.

REPORT: WHAT’S DIGITAL ABOUT FASHION DESIGN?

F.I.R.E and AAM funded by NEMODE Network+ (New Economic Models for the Digital Economy). 

Gabrielle Miller, Mary Jane Edwards, Prof. Sandy Black

What's Digital About Fashion Design?

Interested in Fashion, Technology and the Digital Economy?

We're hosting a roundtable event in partnership with The London College of Fashion and their collaborative platform, FIREup led by Professor Sandy Black entitled - What's Digital about Fashion Design? on Tuesday 27th October 2015.

This event offers insight into current uses of technology within the designer fashion community and a facilitated discussion on the opportunities presented by the digital economy to create new business and economic models.

While the largest global fashion companies are able to invest heavily in digital infrastructure, there is also a pressing need for smaller operators within the designer fashion community – the vast majority of the sector – to innovate and embed digital thinking in their operations, rather than developing products and then working out how to ‘go online’.

If this topic is of interest to you do get it touch with us as we have a few spaces left!

This research workshop is funded by NEMODE Network+ is an initiative under the Research Councils UK (RCUK)’s Digital Economy (DE) research programme to bring together communities to explore new economic models in the Digital Economy.

Mary Jane Edwards

Hey, Big Spender! Using linked data to inform the management of public funding

Earlier this month Andy penned a blog for the London Data Store on the power of open data to inform public funding.

---

Earlier in the summer we supported a short open data project with London South Bank University and the Big Lottery Fund, who recently released nine years of their funding data, totalling some £6bn. The assumption was that this information will be useful for lots of civil society organisations – to see spending priorities by theme and region, for example.

However, we also wanted to see what value it may have for the Lottery itself, so London South Bank University’s Informatics Department used this data to set a task for first-year informatics students, as part of their end-of-year assignment.

The task was to create a dashboard (using IBM Watson Explorer) of value to the Lottery. This very broad brief turned up some interesting results. Walking around the groups as they presented, three main things struck me about the value of even such a short exercise:

New eyes on old data

The Lottery usually – and quite sensibly – present their data on money spent by programme. However, one student group looked at things in an entirely different way.

They noticed that each line of data also contained the full text of the media release that accompanied the announcement of each new funding agreement. They pulled in this text, built a new dictionary around it, and presented the totals as new ‘themes’ (agriculture, health, activities, etc). This of course enabled oversight that cut across a range of programmes.

While much more work would be needed to hone such findings (for example, text references to press releases or use of video might suggest ‘entertainment’ as a category, when the associated programmes are nothing of the sort), but it was instantly apparent how useful such a challenge to existing thinking and assumptions might be. Further similar analysis, for example, might suggest links between programmes that had not previously been identified.

This potential for fresh thinking is part of the reason why the Lottery released this data in the first place, which is a refreshing approach and one that other funders should now emulate.

The power of linked data

It will come as no surprise to learn that the most interesting findings came when the Lottery’s data was overlaid with other data sources. One student group aligned the amounts invested yearly with GDP, for example, and were fascinated to start exploring how the two correlated.

Perhaps the greatest immediate value to the Lottery, however, came from one student group who aligned the Lottery data with constituency boundary data. It emerged that the Lottery are contacted on a regular basis by MPs and other local political influencers wanting to know details of the amount and type of programmes supported in their patch.

Rather than asking a data team to pull off such data and send it to the enquirer, having a platform in place to do such a thing automatically – and perhaps driven by the person making the request – was obviously of interest as it would deliver both an enhanced user experience and a more efficient and cost-effective system.

The power of partnerships

Perhaps the greatest challenge the students faced was in coming up to speed with the focus and aims of the Lottery; they did remarkably well in such a short space of time. This highlighted an area that for me needs more work – the convening of more social sector data partnerships

We need more civil society organisations to find ways to share their data, to provide the basis of useful experiments. The three corners of a potentially powerful data triangle would be:

1. Civil society organisations – who can provide whatever data they may have, highlight which problems are most pressing and add valuable context to the issues in hand.

2. Open data providers such as the London Datastore – whose data can provide many more levels of context to help unearth insights.

3. Universities – who can provide the infrastructure and skills to analyse, explore and find value.

Each group also wins.

The civil society organisations receive data analysis that they could probably not afford. The open data providers get more people looking at and applying their data in socially useful contexts. The universities get real-world problems for their students to grapple with, and more evidence of how they are helping to drive real social impact.

It will be interesting to see how more partnerships such as this are brokered in the coming years.

Many thanks to the excellent Nigel Phillips from LSBU, and the very supportive Simon Marshall from the Big Lottery Fund. Hat-tip also to the energetic360Giving team – and especially Will Perrin who came along to judge the entries – for pushing hard for a wide range of funders to publish their data.

Innovation in employment support: is it just code for saving money?

Earlier this year, in partnership with Nesta and The Centre for Economics and Social Inclusion, we held a collaborative roundtable event exploring existing and emerging innovation supporting those out of work. We invited experts in commissioning, managing, delivering and evaluating employment support provisions to identify the most promising areas for innovations, as well as the barriers that prevent ideas from being tested or scaled.

In this post we’ve shared some of the key points from the session that helped frame the opportunities to develop, and test better models, which deliver better outcomes.

Defining innovation

Early on in the process several participants rightly challenged if there was any ‘real’ innovation occurring at all in employment support. Innovation was often used to describe incremental and procedural changes to existing models and well-established processes, rather than more radical reform. It was felt in the majority of cases, lowering costs and maximising efficiencies rather than producing better outcomes largely drive innovation in this context. Participants also suggested that the current commissioning structures did not incentivise a step-change in thinking or approach, and that commissioners should understand that innovation doesn’t necessarily mean something new.

We were however, encouraged by a number of emerging areas of innovative practice at critical points in the process of supporting someone from unemployment to employment. You can read more about the themes we identified prior to the event here.

Over the course of the event, these themes were further interrogated to include new areas of interest and questions as set out below.

Testing ‘what works’

There was a clear appetite from participants for developing a better understanding of the early indicators of the probability of finding, sustaining and progressing in work. Whilst there are legitimate reasons for the array of different methods utilised to collect evidence there was much questioning as to what extent these measures relate to impact. For example, can they tell us much about the likelihood of an individual finding work and what can they do to help change this prediction?

For this type of approach better tools and processes are required which improve segmentation but also provide more information about the underlying problems that act as a barrier to finding work. Furthermore, it was questioned why, aside from cost, were randomised control trials (RCTs) rarely invested in so standards of evidence could be created which build confidence across the sector. 

Distance travelled and longer-term outcomes

Similarly, more nuanced measurements for ‘distance travelled’ (a person’s progress from unemployment towards employment) would be beneficial in a sector ruled by binaries e.g. in/out of work. It was understood that initial engagement can be a fundamental part of an individual’s journey – long before they are anywhere near a job outcome. However, it was also acknowledged that such systems can act as disincentives for people to be supported into long-term sustainable work.

This thinking extended into in-work support, and the value of longer-term measurements, i.e. is it sustainable, fulfilling, well-paid, secure, are there opportunities for in-work progress and development? There was much interest in innovation and learning around longer-term outcomes that take a more holistic view of progression and earnings, for example Participle’s work around Relational Welfare. It was also considered essential to find a way of addressing everyone in the system, including those who do not secure work post mandatory programmes.  

Data

The discussion around evidence was largely underpinned by data. Recognising that whilst there is a significant amount of data collected, there are no shared systems for analysis or a unified outcomes framework. Successful innovation often relies on information and data sharing to identify promising models, but also to understand performance variation and avoid possible replication of unsuccessful schemes or initiatives. Would providers be more willing to share data publicly if there was greater flexibility and acceptance of failure in contracting? Also, how can a focus on the individual be maintained while making sure it can be modelled, measured, expanded, built upon?

Risk and failure (within commissioning structures)

It is evident that funding for risk and innovation is somewhat scarce in welfare to work. Providers don’t have the resources to risk failure and so have little incentive to innovate. In order to embrace innovation funders, commissioners and delivery organisations need to be comfortable with the down side of innovation and make room for trial and error in programme design. Could there be ring-fenced funds in contracts or dedicated centres for work that support experimentation and learning? Similarly, the specifications of the current commissioning structures and processes should work towards developing a more robust framework for experimentation.

User-led service design  

In scanning the current practice in employment support there was a dearth of approaches akin to design thinking and service user-led design. This arose early on in discussions in relation to the current, binary employment support system that creates a tension in service user power, which is more pronounced than in other services such as health or education.

Given the growing demand for more individual and traditional customer-orientated services, user-led design was considered to be a helpful lens to shift motivation to innovate from improving productivity of services to well-articulated customer need and customer choice. However, a counter point was outlined in that it can be difficult to filter down user-led service design, policy or legislation when it is passed on to delivery managers who generate guidance templates without having been fully involved in generating the evidence – thus making it difficult to test policy or design on a large scale, with varying (often disappointing) results.

Employers

At the demand side of the market there was recognition that whilst employers play a vital role they are not always actively engaged in designing employment support provisions. There were relatively few employer-led models for those farthest from the labour market, and limited examples of joined up engagement across the sector.

Digital

Whilst data collection, analysis and sharing were considered to be integral to furthering the innovation agenda, there was limited reference to the potential of digital technology more broadly. Digital products, platforms and services are currently utilised in back end systems, but how can they add value in terms of user experience, or be embedded within operations, engagement and sector wide systems?

Next steps

It is always difficult to promote innovation as a priority in any area of policy or practice, which is as complex and diverse as employment support. What was however demonstrated even at this initial gathering is the growing desire to stimulate new networks and frameworks for such exploration – and for that exploration to go beyond ‘business as usual’ and only as a measure for ‘saving money.’

Given the diversity and multiplicity of the sector, there’s inevitably a big question around coordination and the roles different actors should play, which challenges the perceived incumbency of large-scale government initiatives and silo working across the sector as a whole. With the drive for more integrated public services and devolution across the country it is imperative that networks of providers exist, and more modular design ensures everyone can do what they do best.

We hope to stimulate new partnerships and activity in this field and further explore:

• The role of data collection and analysis in employment support services to both act as a predictor for finding work and sustaining and progressing in work, but also to demystify key milestones pre-job outcome.
• The needs of employers in the design and delivery of employment support programmes.

If any of the areas above are of interest to you, please do get in touch with Mary Jane Edwards  

Guest post from Southbank University - LSBU informatics students hit the jackpot in assignment for Big Lottery Fund

LSBU's Computer Science and Informatics Department set students a challenge of analysing and interpreting grants data for the Big Lottery Fund.

First-year informatics students have used the latest in analytics technology for their first-year assessments, working with the Big Lottery Fund on a data interpretation challenge.

Students were given a raw set of data showing the dates, amounts and recipients of Big Lottery Fund grants from 2004 onwards. Their brief was to work in groups to create a visual dashboard for the Big Lottery Fund, illustrating the key findings, trends and breakdowns in the data that would provide the organisation with new business intelligence and insight.

Students used IBM's intuitive Watson Analytics technology to prepare their dashboards, which they then presented to a panel of industry experts, including Simon Marshall, Senior Policy and Learning Manager, the Big Lottery Fund, Andy Hamflett, Director, AAM Associates and Matthew Robinson, Solutions Architect at IBM.

Following the dashboard presentations and a student vote, three teams were selected as challenge winners. These and two other high scoring teams will visit the headquarters of the Big Lottery Fund later in the year to present their findings to key staff.

Matthew Robinson, IBM, commented: "What's interesting is how all the groups approached the task in different ways, using their imagination to draw some powerful conclusions. Data science is a big growth area and these students' skills can help businesses make better use of the data they've got, quickly and easily." Andy Hamflett, AAM Associates, continued: "Connecting data science students with real-world problems is great because it encourages fresh thinking," to which Simon Marshall, Big Lottery Fund, added: "I was impressed by how well the students understood what the data meant, they really knew how to join the dots."

The challenge was designed to help students to make connections across all that they had learnt in their first year of study and forms part of the first-year informatics assessment process.