nommon smart cities css bigdata
TRANSCRIPT
Smart Cities, Complex Systems,
Big Data and other stuff
Ricardo Herranz
Co-founder and CEO
CIBALL, 30 April 2013
about us
technology company based in Madrid
founded in 2012
predictive modelling and decision support
“Prediction is difficult, especially about the future”
Niels Bohr
our mission
assist both private and public organisations in creating
sustainable value,
by helping them make better-informed decisions in
complex and uncertain environments
we provide decision support tools and consulting services
for the design, optimisation, and management of
socio-technical systems
socio-technical systems
from policy making to business management, decision-makers deal with socio-technical systems
“Everything should be made as simple as possible, but not simpler”
Albert Einstein
modelling socio-technical systems
the modelling approach must be suitable to reproduce the salient features of the system:
• heterogeneity,
• emergent behaviour,
• self-organisation,
• uncertainty…
traditional, reductionist approaches are not well adapted to model complexity
complex systems science: a new paradigm in socio-technical systems modelling and management
complex systems science
overview and applications
many interacting units
autonomous, decentralised decisions
different spatial and temporal scales
emergence
collective phenomena not reducible to unit behaviour
complex adaptive systems
adaptability: adapt to each other and to the
environment uncertainty
from reductionism…
…to emergence
“We say nothing essential about the cathedral when we speak of its stones”
Antoine de Saint-Exupéry
and of course CITIES
(we’ll talk about it later)
complex systems science
macroscopic (emergent) behaviour
microscopic processes
graph theory
ABM
stochastic processes
multiscale modelling
non-linear dynamics
game theory
from micro to macro
solid state physics
elementary particle physics
the elementary particles of science B obey the rules of science A
yet B is not just “applied A”: new concepts and laws are needed
complexity science is interdisciplinary in nature
mollecular biology
chemistry
cellular biology
mollecular biology
sociology
psychology
macroeconomics
microeconomics
science B: macroscopic (emergent) behaviour
science A: microscopic processes
…
…
…
complex systems science
modelling complex
socio-technical systems
tools and techniques
agent-based modelling
data science
game theory stochastic modelling
network theory modelling of complex
socio-technical systems (STS)
tools & techniques
ABM
• behaviour defined at individual level, global behaviour emerges from agents’ actions and interactions (bottom-up)
• agent features:
– autonomy: acts autonomously without exogenous interventions
– social behaviour: interaction with other agents
– re-activeness: responds to external influences from its environment
– pro-activeness: acts with initiative and goal-orientation
ABM in STS modelling
• include heterogeneity and individualism
• explicit space and local interactions
• bottom‐up analysis: capture emergent phenomena without aprioristic assumptions
• bounded rationality
• direct correspondence with real world entities
• participatory processes in modelling and validation
game theory
• conflict and cooperation between (rational) decision-makers
• players make decisions choosing from a strategy space
• the resulting outcome (payoff) depends on other players’ decisions
• different types of games
– cooperative / non-cooperative
– symmetric / asymmetric
– stochastic games
– perfect / imperfect information (e.g. delayed or partial information)
– games with non-rational players (evolutionary game theory)
game theory in STS modelling
descriptive use:
model how agents’ actually make decisions
(vs prescriptive or normative use:
determine which decisions should be taken)
network theory
a graph is an ordered pair G = (V, E) comprising
– a set V of N vertices (or nodes)
– a set E of K edges (or links, or lines)
undirected graph directed graph weighted (undirected) graph
network theory
adjacency matrix
structural properties
degree distribution
characteristic path length
betweenness
assortativity
clustering
motifs
…
network theory in STS modelling
• understand the networks underlying complex systems
• network topology influences the dynamics of the processes on top of it
– example: shortcuts (small world property) speeds up communication
stochastic processes
probability distributions for the potential outcomes of a problem
random processes:
– discrete: Markov, Bernouilli…
– continuous: Poisson, Wiener (Brownian motion)…
Monte Carlo simulation
stochastic processes in STS modelling
• socio-technical systems are rife with uncertainty
• stochastic nature of many system elements, including agents' decision-making rules
• distributions can be sampled, inferred, or modelled from the set of actual data available
• a comprehensive treatment of uncertainty is a requisite for:
– development of plans that are robust (graceful degradation) and have the greatest likelihood of success
– comprehensive assessment of risks and opportunities
data science and Big Data
• buzzwords that encompass a set of tools and techniques to capture, integrate, manage, analyse and visualise large data sets in order to extract non-trivial, relevant information
• data science includes a variety of techniques from statistical analysis and artificial intelligence
• Big Data:
– from structured to unstructured data
– high volume, high velocity, high variety
data science in STS modelling
• predictive (non-explicative) models
• discover patterns that can suggest new theoretical models, e.g., agents’ behavioural rules (yet, it is worth reminding that correlation is different from causation)
• model calibration and validation
• exploration of the parameter space
some example applications
computational sociology
cultural drift
opinion formation
conflict resolution
segregation
…
organisational intelligence: consumer behaviour
computational economics
epidemiology: disease spreading
source: http://www.gleamviz.org/
traffic modelling
and of course CITIES
(we’ll talk about it in a minute, we’re almost there…)
what we do
“The future cannot be predicted, but futures can be invented”
Dennis Gabor
shaping the future
• focus on prediction
• data mining - analyse the past to extrapolate the future
• lack of explanatory power
business analytics perspective
• focus on fundamental mechanisms
• explanatory models
• predictive power not yet fully exploited in practical applications
complexity science perspective
• integrative approach, exploit the synergies between data science of complex systems science
• understand - model - predict - assess - explain
Nommon perspective
“Don't get involved in partial problems, but always take flight to where there is a free
view over the whole single great problem”
Ludwig Wittgenstein
our vision
• comprehensive decision-making frameworks
• multidimensional nature of the problems holistic approach
• analytical and simulation models
• complex systems science
• statistical analysis, data mining, operations research
quantitative models
• multidisciplinary team: engineering, physics, mathematics, economics…
• open partnership with universities, research centers, industry, public bodies
interdisciplinarity and
multidisciplinarity
• intense R&D activity
• transfer knowledge from academia to policy and business, and ultimately to society
knowledge transfer
databases
solutions
findings: patterns, trends…
model building & calibration
results
validation
findings: patterns, trends…
data mining
data mining
scenarios (strategies, policies…)
KPIs (optimisation, trade-offs, sensitivity analysis…)
models: virtual laboratories
agents’ decisions & interactions
information flow
physical layer
smart cities urban mobility
energy systems
markets
air transport
business intelligence
strategic planning
policy studies
sustainability analysis
demand management
optimisation
products and services
simulation & decision support tools
consulting services
so how does all this relates to
CITIES?
(at last)
After having attended a number of events about smart cities, you will have already
noticed that 9 out of 10 presentations begin with some impressive figures about world urbanisation process and/or a quote from
Jane Jacobs…
…so I’ll skip this part
smart cities, urban modelling, Big Data… (and some other stuff)
urban challenges
urban modelling - a bit of history
smart cities - opportunities for urban planning
our projects: EUNOIA, INSIGHT
conclusions
urban challenges
in the short term, cities are facing the challenge of overcoming the current financial and economic crisis
but cities are also facing other structural and long term challenges
• globalisation: how to combine competitiveness in the global economy with geographical diversity
• environmental sustainability: energy scarcity, emissions (climate change, local air quality), soil sealing
• demographic, social and behavioural changes (migration, aging…)
• new forms of spatial organisation
• social polarisation and segregation
three fundamental, coupled problems
understanding
prediction
governance
the many components of the natural, social, economic, cultural and political urban ecosystems are strongly interwoven, giving rise to complex dynamics which are often difficult to grasp
the limited understanding of urban dynamics makes it difficult to anticipate the impact and unintended consequences of public action
highly distributed, multi-level decision processes and profound impact on a wide variety of stakeholders, often with conflicting and/or contradictory objectives
urban modelling
urban models are mathematical representations of the ‘real world’ —typically implemented through computational simulation tools— that describe, explain, and forecast the behaviour of and interactions between different elements of the urban system
urban models
urban models
• understanding of urban dynamics explanatory role
• virtual experimentation: prediction of the impact of new infrastructures, technologies, or policies
predictive role
• collaborative policy assessment narrative and
deliberative role
a bit of history
• urban planning:
– up to 1950: blueprint planning
– 1950s-1960s: synoptic planning (systems viewpoint, relating objectives to resources and constraints, heavy reliance on quantitative analysis)
– contemporary era: participatory planning, aiming at integrating a plurality of interests and an active public engagement (transactive planning, advocacy planning, bargaining, communicative planning…)
• urban models:
– 1950s: four-stage transport model
– 1960s: CGE models (based on Alonso's bid-choice land use model), spatial interaction (Lowry-type) models, first LUTI models (aggregated, static)
– 1970s-1980s: aggregated, dynamic models (system dynamics), activity-based transport models
– Current trends: disaggregated, dynamic models (CA, ABM…)
cities as complex adaptive systems
• from the image of a city as a ‘mechanistic system’ to that of a ‘living, self-organising system’ that evolves from the bottom up
• urban planning is moving from a centralised, top-down approach to a decentralised, bottom-up perspective
• the role of policy makers and urban planners is that of nurturing positive emergent phenomena and minimising negative emergent properties
current trends
• disaggregation and bottom-up approaches (activity-based and agent-based models)
• coexistence of a variety of models: cellular automata, ad hoc agent-based models of particular sectors such as housing markets or retail choice…
• urban simulation models to be refashioned to deal with new forms of transport and spatial interaction
the smart city
the smart city
• the ‘smart city’ emerged during the last decade as a fusion of ideas about how ICT might improve the functioning of cities:
– efficiency
– competitiveness
– sustainable development
– high quality of life
• initially a very technocentric concept, critical voices soon arose asking for a critical use of ICT and a citizen-centric approach
the smart city: opportunities
• maturity of the smart city concept:
– ICT
– investment in human, social, and environmental capital
• opportunities for improved urban planning:
– big data
– new models and decision support tools
– policy interfaces and participatory governance
big urban data
• open data
• automatic collection of vast amounts of spatio-temporal data
• longitudinal data (versus traditional cross-sectional data)
theoretical advances
• better urban theories
• better predictive models
improved interfaces, visual analytics…
• advanced models more accessible to policy makers
• new ways of citizens’ engagement
our research agenda
our research agenda
• how data from multiple distributed sources can be exploited to understand location, activity and mobility patterns in cities
spatio-temporal data analysis
• improved theoretical models
• integration into state-of-the-art agent-based simulation tools
enhanced urban simulation models
• integration between visualisation and analytical functionalities
information visualisation and visual analytics
• integration across policy areas
• integration of urban simulation into (collaborative) policy making processes
policy and governance
EUNOIA
EUNOIA
• project funded under FP7 ICT Call 8
• focused on urban mobility
• example research questions:
– use of non-conventional data sources (Internet social networks, mobile phone call logs, credit card data) to analyse mobility patterns
– interactions between social networks and travel behaviour
– integration of improved travel behaviour models into MATSim
• case studies: Barcelona, London, Zurich
coordinator
partners
supporting institutions
EUNOIA consortium
INSIGHT
INSIGHT
• proposal submitted to FP7 ICT Call 10
• focused on location models: housing, retail, public services
• example research questions:
– positive/negative synergies between social and economic activities
– impact of the financial crisis
– coupling between short term and long term dynamics
– sustainability indicators based on land use mix/service availability
– integration of improved location models into UrbanSim
• case studies: Barcelona, London, Rotterdam, Madrid
coordinator
partners
INSIGHT consortium
to conclude…
many open research questions
fundamental research questions: a general theory of cities?
model scalability, multiscale aspects, granularity
model calibration and validation
new forms of governance
…
smart cities bring exciting opportunities
availability of an unprecedented amount of data at different temporal and spatial scales
new forms of data analysis and visualisation
new forms of participatory governance
new technologies for citizens’ engagement
…
yet, some precautions…
yet, some precautions…
data is not (always) a substitute for theory
prediction and prescription before explanation can be risky
(“For most applications we don’t need Big Data, but the Big Picture”)
yet, some precautions…
data analysis isn't just about
fancy visualisation
yet, some precautions…
models and data are not a substitute for politics
(“It’s not me, it’s the data”)
yet, some precautions…
models and data are useless if they are not
integrated into governance process
“We have usually thought of city planning as a means whereby the planner’s creative activity could build a system that would satisfy the needs of a populace. Perhaps we should think of city planning as a valuable creative activity in which many members of a community can have the opportunity of participating if we have wits to organize the process that way”
Herbert A. Simon