Using Quantitative Analysis in Support of Military Intelligence P. - - PowerPoint PPT Presentation

using quantitative analysis in support of military
SMART_READER_LITE
LIVE PREVIEW

Using Quantitative Analysis in Support of Military Intelligence P. - - PowerPoint PPT Presentation

Using Quantitative Analysis in Support of Military Intelligence P. Dobias, P. Eles DRDC CORA J. Schroden CNA J. Wanliss Presbyterian College 28th International Symposium on Military Operational Research 29 Aug-2 Sep 2011, UK Outline


slide-1
SLIDE 1

Using Quantitative Analysis in Support of Military Intelligence

  • P. Dobias, P. Eles

DRDC CORA

  • J. Schroden

CNA

  • J. Wanliss

Presbyterian College

28th International Symposium on Military Operational Research 29 Aug-2 Sep 2011, UK

slide-2
SLIDE 2

2

Outline

  • Context
  • Data sources/considerations
  • Traditional methods

– Trends – Seasonality – Forecasting violence – Assessing enemy

  • Fractal nature of conflicts

– Implication of data structure – Multi-fractal forecasting – Current status of research

slide-3
SLIDE 3

3

Context

  • Providing information to enable

mission planning:

– Enemy intent/capabilities – Terrain/Environment – Human terrain, culture, social structure

  • How to conduct assessment in the

environment characterized by:

– Lack of cultural/social/tribal/religious understanding – Insufficient sources of varying reliability – Incoherent and mutually competing enemy groups

slide-4
SLIDE 4

4

Data sources

  • Demographics

Afghan Central Statistical Office collects and disseminates variety

  • f population stats
  • Economics

Many NGO sources provide info such as wheat/sheep prices or power usage

  • Polling

According to some estimates Afg is the most polled country in the world. Kabul group, NGO’s, ISAF, all conduct polls asking a variety of questions

100 200 300 400 500 600 700 800 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 kWh per capita per annum Afghanistan India Pakistan 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 Jan-06 Apr-06 Jul-06 Oct-06 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10 US$/kg International Grain Council Kabul Herat Kandahar Jellalabad

10 20 30 40 50 60 70 80 90 2006 2007 2008 2009 2010 2011 Score Year

Tentative Polling

Q1 Q2

  • Violence Metrics

Collected by security forces, it is

  • ne of the most reliable data

sources around. Most data is stored in CIDNE (replaced JOIIS in 2010)

slide-5
SLIDE 5

5

Concerns about data

  • “One-of” reportings

– Some organization collects data; process not repeated – Impossible to produce trends

  • Changes in collection methodology and timing

– Incoherent and internally inconsistent data – Trends of limited validity

  • Lack of continuity

– Discontinued collection – Data gaps – Limited usefulness of trends

  • Multiple, often conflicting sources
  • Parallel data storage

– All mil data should be in CIDNE – Number of authoritative spreadsheets containing specific info – Difficult correlating of various data

slide-6
SLIDE 6

6

Trends in violence

  • Strong seasonality

– Peaks in July-August – Lowest in December-January – Dips in April due to poppy season

  • Long-term increase
  • Concentrated along Ring-

Road (populated areas)

– Most violence in South and East

slide-7
SLIDE 7

7

Seasonal decomposition

  • Seasonality in Afghanistan

– Annual cycle, difference over 50% – Must be considered when analyzing changes

  • Methodology

Multiplicative model X = T x S – Average X over one season – X/<X> provides raw seasonality, is used to obtain S – T = X / S for each point

  • Long-term trend

Can be used to correlate with factors that do not have seasonal components

  • Assessment

– Identification of recurrent patterns – Identification of long-term trend – Correlations with other factors (friendly activity, weather anomalies) – Deviations from the trend – Implications for the future activities

slide-8
SLIDE 8

8

Use of violent data

  • Understanding enemy

– What is the enemy’s intent? – What are the enemy’s capabilities? – How does the enemy allocate resources? – What is the enemy’s refit/resupply cycle? – How does the enemy adapt to our OPS?

  • Limited value if used

alone; needs supplementary info sources and qualitative analysis

  • Forecasting and risk

assessment

– What violence levels are expected? – Management of resources (medical, materiel, personnel) – Based on assumption that historical trend can be projected to the future – Usually encapsulates some relationship between violence and other factors (e.g. troop numbers, major events)

slide-9
SLIDE 9

9

Assessment of Insurgency

  • What is the state of

insurgency?

– What are their capabilities, intent, morale?

  • Model and Indicators

– Developing a model of insurgency to identify indicators – Combination of violence categories:

  • Effectiveness
  • Particular attack categories
  • Ratios of particular categories
  • Target

– Supported by other sources

  • What are the insurgent

resources?

– How are they distributed? – Origin of resources (local/external)

  • Violence as indicator

– Particular event categories – Distinguish between dedicated and opportunist fighters – Indication of insurgent focus and intent

slide-10
SLIDE 10

10

Forecasting

  • Assumptions:

– Past connection between violence and a factor X will hold – Seasonality will remain the same – Behaviour of factor X

  • Deterministic vs.

stochastic model

– What are other uncertainties? – Is the nature of randomness known? – Are the trials independent? – Is the statistical distribution known or can it be inferred?

slide-11
SLIDE 11

11

Fractal Structure of Violence

  • Power-law

– Fractal nature of the data is reflected in the power law distributions

  • Temporal, Spatial,

Event-based characteristics

  • Persistence

– A result of the memory in the system (the numbers of events at various times not independent) – Implies criticality or near-criticality

slide-12
SLIDE 12

12

Multi-fractal forecasting

  • Identify “trigger”

threshold

– Binary approach (below/above threshold) – Time between crossing threshold (waiting time) – Exploits universality of scaling and persistence

  • Enable short term

forecast:

– More efficient resource allocation – Expectation management – Consequence management

slide-13
SLIDE 13

13

Ongoing activities and future plans

  • Fractal Properties of

Irregular Warfare

– Revisit scaling properties for extended data sets – Revisit intermittency and persistence – Agent-based modeling

  • f small to large scale

combat – Identifying key drivers

  • f fractal behaviour
  • Multi-Fractal

Forecasting

– Revisit persistence of expanded data sets – Test thresholding algorithms – Test multi-fractal forecasting on limited data sets – Test predictive power and validate on real data

slide-14
SLIDE 14

14

Conclusions

  • Quantitative analysis can provide

a different perspective and additional insights into the enemy

  • It cannot be a standalone activity

and needs to be supplemented by qualitative assessments

  • Simple, conventional methods

can provide insights directing further analysis

  • Advanced methods can capitalize
  • n the internal dynamics of

conflicts as complex systems

1 2 3 4 5

slide-15
SLIDE 15