4th Generation Decision Analysis

Lone Star Proprietary Information

4th Generation Decision Analysis


TruNavigatorTM

The Next Generation of Decision Analysis


Lone Star, internationally recognized in the fields of decision analysis, business modeling and simulation is pushing the

state-of-the-art in decision support tools with TruNavigatorTM.


The TruNavigator Decision Analysis tool harnesses the sophistication necessary to enable clients to peer into the future

and gain a true understanding of the range of potential outcomes for any

decision that matters to them, regardless of size or complexity.


With TruNavigator there are now few questions which cannot be answered.


Lone Star wanted to build a new tool from a “clean sheet of paper”

— a tool able to address the most complex DA questions in a highly

scalable format while modeling the options and influences on a laptop

computer.


Evolution of Decision Analysis (DA) Dr. Ron Howard coined the term “decision analysis” in 1965. To paraphrase his seminal work, a

decision is a choice among alternatives, for which the outcome

has some associated uncertainty and for which we have preferences.


Dr. Howard began the work that would lead to his breakthroughs in DA at MIT

in the 1950’s working with Markov chains — a discrete stochastic process

in which the probabilities of occurrence of various future states depend only on the present state of

the system.


Dr. Howard, while working at MIT and Stanford in the 1960’s, evolved a theory supporting evaluation of

multiple potential outcomes in order to make the best possible decision.


Over the course of his work, he expanded on different approaches

including using influence diagrams to help understand problems and

Bayesian statistics to help evaluate outcomes in the face of uncertainty.


Although there is no widely accepted codification of DA methods, when Lone

Star uses the term we mean analysis using Bayesian methods,

based on Dr. Howard’s work.


We organize the history of DA into four generations, each reflecting increasing

sophisticated uses, increasingly capable software, and increasingly powerful

computer hardware.


Dr. Howard’s initial work in DA was ground breaking in three ways:

First, he was able (in theory) to generate any number of possible decision

outcomes.


Second, he introduced the idea of an influence diagram.


Third, he introduced a more Bayesian approach, which was more powerful than pure Markovian mathematics, while at the same time transparent and auditable.


We consider this foundational work to be the 1st generation

of decision analysis.


In the 1980’s Dr. Howard, along with Carl Spetzler, James Matheson and Jeff Foran,

founded Strategic Decisions Group (SDG) to help support organizations with their

decision analysis related requirements.


Dr. Howard and SDG introduced the idea of “decision quality” and integrated

behavioral science into their DA work.

Two of Lone Star’s founders, Volpi and Roemerman,

were SDG clients.


In the course of building the SDG business, they developed basic software which

integrated the theories and approaches espoused by Dr. Howard. Although, the DA

software was limited by the computing power then available, the advancement in capability embodies the 2nd generation in

decision analysis.


As the value of decision analysis became more apparent and

computing power improved and became cheaper,

new 3rd generation tools were built.


Two examples include Chevron’s development of an

internal DA tool to support oil exploration and other critical decisions and the

development of Analytica by Lumina Decision Systems in 1991.


Analytica was one of the first tools marketed as a commercial-off-the-shelf

(COTS) tool in support of DA related requirements. These tools continued to use

many of the premises and approaches promoted by Dr. Howard, while taking

advantage of the improvement in computing capabilities.


4th Generation Decision Analysis

Lone Star’s work with clients such as the Department of Defense,

NASA, AT&T and Hewlett Packard identified limitations in

3rd generation DA tools.


The questions posed by these customers required degrees

of mathematical and organizational complexity

3rd generation models could not address.


In addition, the levels of complexity

led to models of enormous size and the scalability in

3rd generation models was severely limited.


These limitations led Lone Star to build a new tool

from a “clean sheet of paper” — a tool able to address the

most complex DA questions in a highly scalable format while

modeling the options and influences on a laptop computer.


At Lone Star, we believe 4th generation decision analysis tools have the following attributes: Integrated Influence Diagram – to

ensure the relationships are the same in the simulation as in the diagram

Useful Influence Diagram (knowledge capture)


Simple User Interface – no programing or software command structure; intuitive point and click interface with built-in error detection, and editing features

Consistent with cognitive studies – can incorporate subject matter expert opinion in a manner consistent with human cognitive limitations


Math friendly – able to support BOTH statistics experts and the math adverse, capacity to support hard data inputs, data mining inputs, and DIST™ inputs

Black Swan Compatible – able to report on outliers


Computationally Powerful – able to support “what if” excursions with many thousands of Monte Carlo trials in a few seconds — real world problems can require more than 10,000 individual variables which must be represented as uncertainties and executives need to be able to brain storm “what if” questions and see answers in seconds


Lone Star’s TruNavigatorTM represents4th generation decision analysis


Why It Works

The long answer is the tool is based on sound mathematics, solid code development, and evolution

through empirical use in hundreds of client engagements.


GUI helps users quickly build an influence diagram which clearly shows decision options and

contributing factors


DA is about predicting outcomes. Typically, when you hear of a prediction,

the answer is given as a single deterministic value.

When dealing with future outcomes, this approach is fundamentally

flawed and seriously misleading.


Rather, predictions need to be thought of as clouds consisting of a

range of probable outcomes. he ability to provide quantitative insight

on the likelihood and consequences of each outcome makes

TruNavigator useful and is the core of what we do.


At the most basic level, Lone Star decomposes a

customer’s decision or problem —the Big Question —

into all the elements which influence it. Data is collected to describe each element.


Using built-in Monte Carlo simulations,

Bayesian mathematics and other proprietary calculations,

TruNavigatorTM predicts unknown elements based on related influences,

tests customers’ decisions and identifies probable future outcomes.


TruNavigatorTM works backward from the Big Question.

Our process is based on learning from domain experts at all levels within the

customer’s organization. We ask a series of “Little Questions” about their outcome

expectations for the events which make up the influences on the Big Question.


This process is less intimidating than requiring an answer for the

Big Question and provides a deeper understanding of the interrelationships within the organization.


TruNavigatorTM can generate up to 50,000 Monte Carlo sets per model node

in just a few seconds, on a laptop computer. Using a large number of these calculations is a fundamental requirement to increase

the confidence level for management’s decisions by allowing a review of

multiple “what if” questions and/or related scenarios.


The entire model is built using the language of the

customer’s organization, not predefined Lone Star terms,

which makes interpreting the model’s outputs more easily understood and useful to the organization.


The graphical outputs of the model provide an intuitive way to see the effects of

potential decisions. Sensitivity analysis of any strategic element can be run to

determine what influences it, what matters most in determining the outcome, and

which outcomes have the best profitability or the greatest risk.


The analysis process typically promotes running an infinite number

of “What-if” scenarios, which can be built quickly

with results generated in minutes.


TruNavigatorTM is being used by large and small organizations, including defense agencies and

prime contractors as well as enterprise

and entrepreneurial ventures.


TruNavigatorTM Benefits

Universally customers tell us that in addition to the accuracy TruNavigatorTM delivers, an

unexpected benefit is increased consensus within the management and project teams

in making the final decision.


Additionally, clients found they gained a better understanding

of interrelationships across their organization.


The advantages of TruNavigator™ are many, including: Equips executives and managers to

effectively manage uncertainty Allows an organization to assess the

impact of decisions before resources are committed


Evaluates decisions/programs/systems in the context of complexity and interrelationships

Provides a GUI interface for easy development of the model and fast additions at any time


Supports graphical output of results yielding an easy way to see the effects of potential decisions

Offers sensitivity analysis of any element of the strategies being considered

Uses the language of the customer’s organization, not predefined LONE STAR terms


Improves comprehensiveness of considered decision option through “What-if” scenarios which can be quickly built “on-the-fly” with results seen in minutes.


About Lone Star Headquartered in Dallas, Texas, Lone Star provides business and technical analysis and advisory services addressing clients’

most complex, mission critical challenges. Using processes based on best practices

and proprietary tools, Lone Star delivers effective outcomes and

the value clients expect.


Lone Star’s unique knowledge and analytical capabilities and a client service-

oriented philosophy means actionable results are always delivered. Lone Star’s

roots lie in the development, fielding and support of complex technologies and

programs for the Department of Defense and leading enterprises.


For more information,additional white papers

www.lone-star.com or contact:Matthew Bowers, VP Corporate Development

214-440-3184 [email protected]

http://www.lone-star.com/

4th Generation Decision Analysis

Business

Transcript of 4th Generation Decision Analysis