Tim Stonor_Predictive analytics_Kyoto Smart City Expo presentation
1. Predictive analytics
For urban planning, building
design & spatial economics
Smart City Expo 2014
Kyoto
26th March 2014
Tim Stonor
Architect & Urban Planner
45. People
1. Behaviour
2. Wealth
3. Health
4. Safety
5. Cultural identity
Resources
1. Materials
2. Energy
3. Finance
4. Utilities supply
5. Waste handling
Urban Form
1. Spatial Form
2. Physical Form
- capacity
- location
- condition
- value
dActions
v3
dDecision-taking
Environment
1. Sun
2. Wind
3. Rainfall
4. Groundform
5. Biodiversity
Data Mapping
Data Quantification
Visualise Urban Data Collective
Integrated platform for performance data
How is the place working?
dUncertainty
dIdeas for change
Opportunities & Constraints
Creation of Conceptual Options
Conceive Urban Strategy
Integrated urban policy and practice
What are the opportunities for change?
dConsultation
Forecast Urban Forecast Model
Scenario testing & performance forecasting
How will performance be affected by change?
Simulation & Optimisation of Options
Development of Preferred Option
dAwareness
Explain Urban Performance Model
Integrated urban diagnostics: spatial analytics
Why does the place work the way it does?
Urban Diagnosis
Issues, Objectives & Principles
Urban Baseline Study
Urban Impact Analysis
User interface/state
Behaviour change
Decision groups
Data
Data inputs
1
2
3
4
5
6
Integrated
Urban
Model
Good afternoon. In this presentation I will show you an approach to urban planning that on computer-based predictive analytics. As an architect in private practice, these tools are the foundation of my company’s approach: one that uses a science-based approach to understand human behaviour in cities. Tomorrow I will be speaking again about the Foresight work that the UK Government Office for Science is undertaking to address the Future of Cities, a project that I am contributing to, coming from the background that I would like to now take you through.
I think it’s important that we start with some of the challenges facing current urban planning practice worldwide.
Well all of that sounded terribly depressing. So how can we address these challenges? Well, starting with the behaviour of people is a good approach. Here on the left is the Tate Gallery in London and we have followed 100 people for 10 minutes, picking them up at the main entrance at the bottom of the image, then tracking them. You can see their individual traces as well as the overall pattern they make. Lots of people walk up the central axis of the gallery. More walk on the left than the right sides.Now on the right is a spatial analysis – quite a simple one – where a computer breaks the plan of the building down into discrete elements of space – you can perhaps see the individual square pixels…
SkyCycle is born of an interest in how the city connects. My company, Space Syntax, has created a unique approach to Spatial Network Analytics. We predict the movement of people in cars, on foot and on bikes and, as with SkyCycle, our work is often about spotting strategic opportunities to make cities work better for people.
Which gives us a definition of the city as a transaction machine, built on its spatial network – indeed very largely dependent on its spatial network to organise flows of people, goods, energy and communications. After all, it is largely along streets that pipes, cables and tunnels all run. This makes the management of existing spatial networks and the creation of new ones a key element of infrastructure planning.
A brief pause to recognise the central role that University College London has played in the development of this technology, supported by funding from the UK Research Councils. The tools I have shown you are based on over 30 years of continuous development, which continues to this day.
MAPSpace Syntax’s cycle network analysis begins by mapping existing cycle infrastructure to evaluate network capacity, legibility and character. This approach is underpinned by a well-established method of analysing the wider spatial layout hierarchy as well as the location of transport nodes and other attractor land-uses which influence cycling behaviour.MeasurePatterns of vehicle and pedestrian movement are also examined and conflicts are studied. Spatial and temporal patterns in movement are identified and analysed.Model Ideas for change are then modelled to evaluate their impact.Space Syntax models create a unique understanding of how physical and spatial factors interact to influence the way that all roads users – including cyclists and pedestrians – move, interact and transact in streets.MakeSpace Syntax’s models are used to highlight key issues and predict future demands in order to identify what and where investment is needed.Applications include evidence-based cycle policy and strategic network development as part of integrated, multi-scale and multi-modal public realm design concepts.
These tools are increasingly being used internationally. Recently we launched a vision for the expansion of the Central Business District in Darwin, Australia, where the predictive analytics were used to identify optimal pattern of new streets and public spaces, as well as a distribution of land uses. Key to this was modelling of land value impact, both on the new development and on existing land.
The evidence base that we have created has proved useful in stakeholder negotiations.
But always with an eye on the human scale of local behaviour and interaction – as here in a plan for the improvement of an unplanned area of Jeddah – where we have created a method for better integrating the informal development into the wider city. This is both a spatial and an economic strategy.
And predictive modelling is proving highly useful in the creation of large and rapid urban expansion plans, as here in the north-eastern Chinese city of Changchun, where you can see the very large growth that is planned.
So, in summary, what I have show you is an urban modelling system, created in the UK but now being used and developed worldwide by well over 5,000 different people. It brings together data sets from various sources and the key to its operational effectiveness is its Spatial Layout Analysis – the base layer in the data stack.
This Integrated Urban Modelling approach addresses those urban challenges I listed at the start of this presentation. It takes different kinds of data inputs across the top of the page. Then by first Visualising, then Explaining data correlations it allows urban planning ideas to be Conceived and their impacts to be Forecast. My experience has been that this takes people from a state of Uncertainty, through Awareness of key issues, to become contributors themselves of Ideas for Change, all of which aids in the Consultation and Decision-taking processes. With the ultimate aim being to take actions that lead to behaviour changes in pursuit of higher order goals of social cohesion, economic productivity and carbon reduction.
So if we look again at the challenges, my view is that predictive analytics can address each of them, allowing ultimately for decisions to be taken that might seem bold but are actually risk reduced.