system dynamics, analytics & big data (16th conference of the uk chapter of the system dynamics...
DESCRIPTION
This talk investigates the relationship between system dynamics, analytics and big data. Drawing on both a historical analysis and text analytics, similarities and differences are identified, and some suggestions on how future research may provide value for the System Dynamics community.TRANSCRIPT
Analysing Analytics:Evolution or Emperor's New Clothes?
System Dynamics Society, April 2014Michael Mortenson, Neil F. Doherty & Stewart Robinson
Big Data, Analytics & System Dynamics:
The Red Pill or the Blue Pill?
Structure1. Background
2. Relationship Problems
3. The Dianoetic Management Paradigm
4. Categories of Analytics
5. Implications for System Dynamics
2 Big Data, Analytics & System Dynamics – April 2014
Competing on Analytics
3 Big Data, Analytics & System Dynamics – April 2014
190,000shortage of analytics specialists in the US
alone (Manyika et al, 2010)
$225,000starting salaries for data scientists
(Loizos, 2013)
$300p/hhourly rate to hire data scientists
via Kaggle (Granville, 2013)
1. Why Analytics?
Big Data, Analytics & System Dynamics – April 2014
$105,000,000,000size of the business analytics market in 2010 (IBM, 2010)
83%“of c-suite executives agree the importance of using information effectively has never been
greater” (SAS, 2009)
4
1. Why Big Data?
0
200,000,000,000,000
400,000,000,000,000
600,000,000,000,000
800,000,000,000,000
1,000,000,000,000,000
1,200,000,000,000,000
3,000,000,000,000
1,200,000,000,000,000
How Much Data is There in the World?
40,000%
2010
1997
Sources: Lesk (1997) and Gow (2010) Big Data, Analytics & System Dynamics – April 20145
1. Analytics & Operational Research?
Big Data, Analytics & System Dynamics – April 2014
The Analytics Networkwww.theorsociety.com/Pages/SpecialInterest/AnalyticsNetwork.aspx
6
1. Big Data & System Dynamics?
Big Data, Analytics & System Dynamics – April 20147
1. The Red Pill or the Blue Pill?
Big Data, Analytics & System Dynamics – April 20148
2. Relationship Problems
Big Data, Analytics & System Dynamics – April 2014
≈Analytics OR/MS
Analytics
OR/MS Analytics
OR/MSOR/MSAnalytics
≠Analytics OR/MS
6% 7%
28% 29% 30%Source: Liberatore and Luo (2011)9
2. Relationship Problems
Big Data, Analytics & System Dynamics – April 2014
vs. vs.
10
3. The Dianoetic Management Paradigm
Big Data, Analytics & System Dynamics – April 201411
3. The Dianoetic Management Paradigm
Big Data, Analytics & System Dynamics – April 201412
System Dynamics
3. The Dianoetic Management Paradigm
Big Data, Analytics & System Dynamics – April 2014
Scientific Management (1910-1945)
Technology1945 Design of the von Neumann Architecture the computer structures still used today1952 The UNIVAC computer predicts the US presidential election1957 FORTRAN programming language devised
Quantitative Methods1947 Linear programming developedc1947 OR/MS methods used to help rebuild UK
industry (Kirby, 2003, pp 190-105)
Decision Making1946 Formation of the Ergonomics Society1947 Simon’s Administrative Behavior published c1959 Judy Clap leads the development of the
first graphical user interface (Grer, 2002)
Technologyc1913 The Ford Model 1 began production using its influential assembly lines1914 The end of The Technological Revolution1941 The first digital computer, Z1, released
Quantitative Methods1935 Publication of Fisher’s The Design of
Experiments1938 First discussions of ‘OR’ (Kirby, 2003 p 71)1939 Development of cluster analysis
Decision Making1912 The principles of Gestalt visual perception
devised (Wagemans et al, 2012)1921 Launch of the Cambridge Psychological
Laboratory designed to distribute the results of studies amongst industry
The Scientific Method (1945-1960s)
13
3. The Dianoetic Management Paradigm
Big Data, Analytics & System Dynamics – April 2014
Management Info Systems (1960s-1970s) Decision Support Systems (1970s-1980s)
Technologyc1963 The development of microchips1964 Release of the IBM System/360c1970 E. F. Cobb conceptualises the first relational databases (Date, 2000)
Quantitative Methodsc1963 Geography’s Quantitative Revolution
demonstrating the growth of quantitative methods in academia (Burton, 1963)
1964 The first UK master’s degree in OR/MS
Decision Making1962 The Myers Briggs Type Indicator published,
used to understand decision maker typesc1962 Behavioural science grows in influence,
particularly in consumer researchc1969 First study into computer-aided decision
making (Ferguson and Jones, 1969)
Technologyc1972 Personal computers are popularised in businesses (Ceruzzi, 1999, pp 207-241)c1972 TCP / IP internet protocols introduced1973 IBM 3660 Supermarket System released introducing barcode scanners
Quantitative Methodsc1975 ‘S’ statistical language and Matlab are
launched. SPSS and SAS grow in popularity (Wegman et al, 1997)
1979 Development of the ID3 decision tree algorithm (the predecessor of C4.5)
Decision Making1979 Research into decision making needs of
CEOs leads to the design of Executive Information Systems (Rockart, 1979)
1981 Development of soft systems methodology
14
3. The Dianoetic Management Paradigm
Big Data, Analytics & System Dynamics – April 2014
Business Intelligence (1980s-1990s) Analytics (2000 – Present Day)Technology1988 The conceptualisation of data warehouse architecture Devlin and Murphy, 1988)1989 Launch of the world-wide-web
Quantitative Methodsc1988 The first significant research into agent
based modelling (Samuelson, 2000)1989 Piatesky-Sharpio introduces the term ‘data
mining’ (He, 2009)c1996 General Electric introduces Six Sigma to its
operations (Henderson and Evans, 2000)
Decision Making1992 Development of balanced scorecards
(Kaplan and Norton, 1992)2000 Popularisation of business dashboards
(Marcus, 2006)
Technology2004 Google’s Dean and Ghemawat publish a paper detailing MapReduce, the big data programming paradigm2004 Launch of Facebook (Twitter in 2006)2007 Development of NoSQL databases
Quantitative Methods2001 The release of the Natural Language
Toolkit, helping popularise text mining2008 Anderson’s The End of Theory published2010 The first Kaggle competition
Decision Making2005 eBay buy shopping.com, highlighting the
importance of recommendation agents2013 Tableau, the data visualisation software,
valued at $2bil after two days on the Stock Exchange (Cook, 2013)
15
3. The Dianoetic Management Paradigm
Big Data, Analytics & System Dynamics – April 2014
The Isolationist Approach
vs.The Faddist Approach
16 Source: Mortenson, Doherty, Robinson (Forthcoming)
4. Categories of Analytics
Big Data, Analytics & System Dynamics – April 2014Source: Blackett, 201217
4. Categories of Analytics
Big Data, Analytics & System Dynamics – April 201418
4. Categories of Analytics
Big Data, Analytics & System Dynamics – April 2014
Descriptive Analytics
Predictive Analytics Prescriptive Analytics
Statistical and data modelling techniques designed to describe past events and answer “what happened”?
Data mining and machine learning techniques used to
predict future events and answer “what will happen next”?
OR/MS, mathematical and statistical models used to prescribe future actions and answer “what
should we do next”?
Technological Strategic
Lower Risk Decisions Higher Risk Decisions
Discovery Analytics Decision Analytics
Advanced Discovery Analytics
Reporting & alertsMarket research
ERP & information systems
Basic historical analysisPerformance metrics
Stakeholder consultation
Advanced visualisationReal time insights
Automated learning models
Advanced Decision Analytics
OptimisationProblem structuring
Modelling & simulation
Advanced
19
4. Categories of Analytics
Big Data, Analytics & System Dynamics – April 201420
Discovery Analytics Decision Analytics
Describe and summarise the data and business context
Describe and summarise the problem situation and/or system
Build models than can make predictions about unseen data
(holdout or future data)
Build models than can predict how the system would respond to
different stimuli or conditions
Prescribe future actions based upon the model
RecommendPrescribe future actions based
upon the model
Recommend
5. Implications for System Dynamics
Big Data, Analytics & System Dynamics – April 201421
5. Implications for System Dynamics
Big Data, Analytics & System Dynamics – April 201422
High volume
data
Unstructured data Streaming &
real-time data
Big data architecture
(e.g. Hadoop)
Data visualisation
Decision automation