3. Complex Systems
“In general usage, complexity
tends to be used to
characterize something with
many parts in intricate
arrangement.”
EF Legara • @eflegara • Asian Inst. of Management
6. Complex Systems in Nature
EF Legara • @eflegara • Asian Inst. of Management
7. Complex Systems in Society
EF Legara • @eflegara • Asian Inst. of Management
8. Complex Systems
• LARGE and DIVERSE
INTERACTING parts
• No centralised control
• SELF-ORGANISATION and
EMERGENCE
• EVOLUTION and ADAPTATION
EF Legara • @eflegara • Asian Inst. of Management
9. Why Study Complex Systems
1. Demographic change
2. Financial and economic stability
3. Social, economic and political inclusion
4. Public health (epidemics)
5. Balance of power and conflict
6. Corruption and crime
7. Collective social behavior
8. Institutional design
9. Sustainable use of resources
10. Reliability of critical infrastructures
Real-World Issues
EF Legara • @eflegara • Asian Inst. of Management
10. “I think the next
century will be
the century of
Complexity.”
14. Bus Arrivals
Waiting time ½ x headway2 =
the area of each
triangle
time
headway
headway
EWT
SWT
AWT
An interactive visual, demand modelling, and decision-support tool.!An interactive visual, demand modelling, and decision-support tool.!
Bus Arrivals
Waiting time ½ x headway2 =
the area of each
triangle
time
headway
headway
EWT
SWT
AWT
Modeling and
Simulations of
the Rapid Transit
System
Reliability
Analysis of Bus
Arrivals
Lightless
intersection
control numerical
simulations
Land-Use &
Transport Modeling
Crowd Modeling
and Simulations
Characterizing
Public Transport
Commuters
Resilience of
Commuter
Encounter
Networks
Aging, Biology &
Computing:
Healthspan
Identification of
Regulators in a
Human Gene
Network
Urban Morphology
Dynamical Model of
Twitter Activity
Profiles
Diffusion &
Cascading Failures
on Multiplex
Networks
Urban
Complexity
Projects
17. Multilevel Marketing
or Pyramid Scams?
A mechanism-based model to
reconstruct network marketing
EF Legara • @eflegara • Asian Inst. of Management
18. Agent-Based Modeling + Network Science
Most real-world systems can only be modeled algorithmically.
“Social Network”
EF Legara • @eflegara • Asian Inst. of Management
19. Agent-based modeling
a class of computational models
for simulating the
actions and interactions of
autonomous agents
EF Legara • @eflegara • Asian Inst. of Management
20. Agent-based modeling
• Bottom-up approach
• Simulate actions and interactions of
autonomous agents in a synthetic
virtual “world”
• Best for “what-if” scenario modeling
Applications
• Epidemiology and Contagion
• Traffic
• Financial markets
** EF Legara (2014). Complexity in Public Policy: Tools and Applications (Complex Networks and Agent-Based Models: An Overview).
Managing Complexity from a Public Policy Perspective (A Lecture), Civil Service College, Singapore, 15-16 October 2014.
ABM in Public Policy**
Organisation for Economic Co-operation and
Development (OECD)
Global Science Forum
21. Agent-Based Modeling
EF Legara, KK Lee, GG Hung, and C Monterola, "Mechanism-based model of a mass rapid transit
system: A perspective," Int. J. Mod. Phys. Conf. Ser. 36, 1560011 (2015).
N Othman, EF Legara, V Selvam, and C Monterola, "A Data-Driven Agent-Based Model of
Congestion and Scaling Dynamics of Rapid Transit Systems," J of Computational Science (2015).
Integrated Public Transportation Visualization and
Simulation Platform of Singapore
Predicting the H1N1 Epidemic
The predicted spread of the H1N1 epidemics during 2009, representing the first successful real-time prediction of a
pandemic [13]. The project, relying on data describing the structure and the dynamics of the worldwide transportation
network, foresaw that H1N1 will peak out in October 2009, in contrast with the expected January-February peak of
influenza. This meant that the vaccines timed for November 2009 were too late, eventually having little impact on the
outcome of the epdemic. The success of this project shows the power of network science in facilitating advances in areas
of key importance for humanity.
Video courtesy of Alessandro Vespignani. Network Science by Albert-Laszlo Barabasi.
EF Legara • @eflegara • Asian Inst. of Management
30. EF Legara • @eflegara • Asian Inst. of Management
31. Aric Hagberg Dan Schult Pieter Swart
The
community!!
EF Legara • @eflegara • Asian Inst. of Management
32. Watts-Strogatz Social Network Model
Unilevel Business Network
Binary Level Business Network
EF Legara • @eflegara • Asian Inst. of Management
EF Legara, C Monterola, DE Juanico, M Litong-Palima, C. Saloma, "Earning potential in
multilevel marketing enterprises," Physica A 387, pp. 4889-4895 (2008).
33. Contrary to claims of unbridled growth.
analytic
actual
simulated
NumberofAgents
Level
https://tinyurl.com/erika-aim-mlm EF Legara, C Monterola, DE Juanico, M Litong-Palima, C. Saloma, "Earning potential in
multilevel marketing enterprises," Physica A 387, pp. 4889-4895 (2008).
38. EF Legara • @eflegara • Asian Inst. of Management
39.
40. GBM
DRF
SVM
Train the
ML models
Adult
Child
Senior Citizen
EF Legara and C Monterola,
"Inferring Passenger Types from
Commuter Eigentravel Matrices,"
Transportmetrica B: Transport
Dynamics, Feb 2017. Doi:
10.1080/21680566.2017.1291377
(“travel DNA”)
44. python
IEEE SCALE Challenge. First Prize Winner
Perfect for prototyping.
python
Scalable complex system modeling
for a sustainable city
EF Legara • @eflegara • Asian Inst. of Management
45. EF Legara • @eflegara • Asian Inst. of Management
47. Why python?
It’s simple and dynamic!
The built-in libraries!
The community!
The reproducibility!
And many more…
EF Legara • @eflegara • Asian Inst. of Management
48. Why Python in
Scientific
Research?
Erika Fille Legara, Ph.D.
Program Director, M.S. in Data Science
Asian Institute of Management
@eflegara | eflegara.github.io