Agent-Based Modeling as a Foundation of Big Data The 9th International Workshop on Agent-Based Approaches in Economics and Social Complex Systems (AESCS 2015), Ramada Bintang Bali Resort, Bali, Indonesia September 9-11, 2015 Shu-Heng Chen,
[email protected] Ragupathy Venkatachalam,
[email protected]
AI-Econ Research Center Department of Economics National Chengchi University Taipei, Taiwan http://www.aiecon.org/
Outline What is Big Data? Two Pioneering Demonstrations What do we Mean by `Foundation’? Concluding Messages
Definition Big Data
Technological Concerns
Psychological Concern
United Nations Global Pulse Physical Concern
A Myriad of Definitions Ward J, Barker A (2013) Undefined by data: A survey of big data definitions. arXiv preprint arXiv:1309.5821. Volume, Velocity, Variety and Veracity
Global Pulse, United Nations
Global Pulse, United Nations
What is Big Data?
Big data is, in a given spatio-temporal domain, the archive of whatever people said, people did and even people thought.
Basically, it is the microscopic view of human activities.
Notebook, Smart Phone, Apps
Notebook, Smart Phone, Apps
Platform Notebook, Smart Phone, Apps
Notebook, Smart Phone, Apps
Examples Google Flu Trends Global Pulse Google Glass Street Bump
Platform
Examples BinCam Environmental Teapot Smart Mirrors Smart Carpet Smart Belt
Two Studies on Swarm
School of Fish
Partridge B (1981) Internal dynamics and the interrelations of fish in schools. Journal of Comparative Physiology 144(3): 313-325.
Flock of Birds
Reynolds C (1987) Flocks, herds, and schools: A distributed behavioral model. Computer Graphics 21(4): 25-34.
Experiments
Data Mining
Agent-Based Simulation
Data
Big Data
Behavioral Rules
Big Data
Marking (Tagging)
Filming
Behavioral Rules: AAA Saithe match changes in both swimming direction and speed of their neighbors but correlations are greater for swimming speed.... Saithe simultaneously match the headings and swimming speeds of at least their first two nearest neighbors within the school. (Ibid, p. 313) Avoidance, Attraction, Alignment
Fundamentalists and Chartists
Frankel J. Froot K (1986) The dollar as a speculative bubble: a tale of fundamentalists and chartists. Technical Report 1845, NBER, Cambridge, MA. (The pioneering paper on the fundamentalist-andchartist model.) Frankel J. Froot K (1990) Chartists, fundamentalists, and trading in the foreign exchange market. American Economic Review 80, 181– 186. (This paper provides the empirical basis of the fundamentalistand-chartist model). Allen H, Taylor M (1990), Charts, noise and fundamentals in the London foreign exchange market, Economic Journal, 100: 49-59. Barber, B., and T. Odean (2000), Trading is hazardous to your wealth: The common stock investment performance of individual investors, Journal of Finance 55: 773-806
Craig Reynolds
What Foundation Means? What are the expected properties of big data? What is the quality of big data? Is big data necessarily satisfied with `wisdom of crowds’? What is information aggregation efficiency which we may have from a specific set of big data?
Foundation of Big Data
The agent-based model enables us to answer the question regarding the quality of big data, quality in the sense of information aggregation efficiency. Based on the characteristics of an agent-based model, such as
number of agents, learning or meta-learning behavior, network structure, personal traits, preferences, and cultures (social norms),
we may articulate the information aggregation process of big data and their inherited properties. This is what we mean by being the `foundation' of big data.
Some Prototypical Examples Prediction Markets Sentiment Analysis in Financial Markets Online reviews (Vriend, 2002)
Agent-Based Prediction Market
Prediction Market
Big Data
Big Data
Limited Order Book Information. Price, Spread, Volatility, Depth, Order Aggressiveness
http://xfuture.org
2014 Annual Meetings of the Eastern Economic Association, Boston, MA, March 6-9, 2014
Containing Capacity (Tolerance Level) =0.75, 2012
Blue = 51.60% Green= 45.63% Orange = 2.77% Grid Size =193193 Population Size = 13,454 Density = 36.12%
Containing Capacity (Tolerance Level) =0.25, 2012
Blue = 51.60% Green= 45.63% Orange = 2.77% Grid Size =193193 Population Size = 13,454 Density = 36.12%
Distribution of Reservation Prices
Demand and Supply Curve
Artificial Price Series
Link One
Social Networks
Price Volatility
Big Data
News
Link Three
Social Medium Networks
Link Two
Concluding Messages
We propose a process-based definition of big data, to be distinguished from size-based, technology-based, or psychology-based definitions. We argue that our definition provides a sharp ontology as well as a new epistemology of big data. The agent-based simulation society must take their data more seriously than its current state. The agent-based model cannot be fully harnessed without the resort to big data analytics.