mgs3100 general modeling chapter 1: introduction
TRANSCRIPT
MGS3100General Modeling
Chapter 1: Introduction
THE MODELING PROCESS
Managerial Approach to Decision MakingManager analyzes situation (alternatives)
Makes decision toresolve conflict
Decisions are implemented
Consequences of decision
These stepsUse
SpreadsheetModeling
A Detailed View of the Modeling Process1. Diagnose the problem2. Select relevant aspects of reality3. Organize the facts; identify the assumptions, objectives, and
decisions to be made4. Select the methodology5. Construct the model6. Solve the model (generate alternatives)7. Interpret results (in “lay” terms!)8. Validate the model (does it work correctly?)9. Do sensitivity analysis (does the solution change?)10. Implement the solution11. Monitor results
THE MODELING PROCESS
ManagementSituation
Decisions
ModelAnalysis
Results
Intuition
Ab
stra
ctio
n
Inte
rpre
tati
on
Real World
Symbolic World
The Modeling Process
ManagementSituation
Decisions
Model
Analysis
Results
Intuition
Ab
stra
ctio
n
Inte
rpre
tati
on
Real World
Symbolic World Managerial
Judgment
Reasons for Using Models
Models force you to:Be explicit about your objectivesThink carefully about variables to include and
their definitions in terms that are quantifiable Identify and record the decisions that
influence those objectives Identify and record interactions and trade-offs
among those decisions
Reasons (cont.)
Consider what data are pertinent for quantification of those variables and determining their interactions
Recognize constraints (limitations) on the values that those quantified variables may assume
Allow communication of your ideas and understanding to facilitate teamwork
Types of Models
Building Models
PerformanceMeasure(s)
Decisions(Controllable)
Parameters(Uncontrollable)E
xoge
nous
Var
i abl
es
ModelConsequence Variables
EndogenousV
ariables
The “Black Box” View of a Model
MODELING VARIABLES
Modeling TermManagement
Lingo Formal Definition Example
Decision Variable Lever Controllable Exogenous Investment Input Quantity Amount
Parameter Gauge Uncontrollable Exogenous Interest Rate Input Quantity
Consequence Outcome Endogenous Output Commissions Variable Variable Paid
Performance Yardstick Endogenous Variable Return on Measure Used for Evaluation Investment
(Objective Function Value)
Examples of Decision Model Assumptions - Profit Models
If it is beyond your control, do not consider it!Overhead costs - a convenient fiction - we
ignoreSunk costs - we ignoreDepreciation - only include if we can use to
shield future taxes Costs are linear in the short term
Building Models
Symbolic Model Construction
Mathematical relationships are developed from data. Graphing the variables may help define the relationship.
Var. X
Var
. Y
Cost A
Cost BA + B
Modeling with Data
Consider the following data. Graphs are created to view any relationship(s) between the variables. This is the first step in formulating the equations in the model.
Creating the Symbolic ModelPredicting Sales Based on Marketing Expenditures
y = 3.5853x + 357.7
R2 = 0.9316
0
500
1000
1500
2000
2500
3000
0 100 200 300 400 500 600 700
Marketing Expenses (x)
Sale
s Re
venu
e (y
)
DETERMINISTIC ANDPROBABILISTIC MODELS
Deterministic Models
are models in which all relevant data are assumed to be known with certainty.
can handle complex situations with many decisions and constraints.are very useful when there are few uncontrolled model inputsthat are uncertain.
are useful for a variety of management problems.
allow for managerial interpretation of results.
will help develop your ability to formulate models in general.
Probabilistic (Stochastic) Models
are models in which some inputs to the model are not known with certainty.
uncertainty is incorporated via probabilities on these “random” variables.
often used for strategic decision making involving an organization’s relationship to its environment.
very useful when there are only a few uncertain model inputs and few or no constraints.
DETERMINISTIC ANDPROBABILISTIC MODELS
Deductive Modelingfocuses on the variables themselves before data are collected.variables are interrelated based on assumptions about algebraic relationships and values of the parameters.
focuses on the variables as reflected in existing data collections.
tends to be “data poor” initially.
Inferential Modeling
variables are interrelated based on an analysis of data to determine relationships and to estimate values of parameters.
available data need to be accurate and readily available.tends to be “data rich” initially.
places importance on modeler’s prior knowledge and judgments of both mathematical relationships and data values.
ITERATIVE MODEL BUILDING
ITERATIVE MODEL BUILDING
DEDUCTIVE MODELING
INFERENTIAL MODELING
PROBABILISTICMODELS
DETERMINISTICMODELS
Model Building
Process
Models
ModelsModels
ModelsDecision M
odeling
(‘What I
f?’ P
rojectio
ns, Decisio
n
Analysis, Decisio
n Tre
es, Queuin
g) Decision Modeling
(‘What If?’ Projections,
Optimization)
Data Analysis
(Forecasting, Simulation
Analysis, Statistical Analysis,
Parameter Estimation)
Data A
nalysis
(Data
Base Q
uery,
Param
eter E
valuatio
n
Philosophy of Modeling
RealismA model is valuable if you make better
decisions when you use it than when you don’t.
IntuitionA manager’s intuition arbitrates the content of
the abstraction, resulting model, analysis, and the relevance and interpretation of the results.
MGS3100General Modeling
Chapter 11: Implementation
Just as knowledge of Excel is insufficient Just as knowledge of Excel is insufficient without modeling concepts, your without modeling concepts, your knowledge of spreadsheet modeling alone knowledge of spreadsheet modeling alone is insufficient for truly affecting decision is insufficient for truly affecting decision making in organizations.making in organizations.
INTRODUCTIONINTRODUCTION
Creating a model itself, although an Creating a model itself, although an important first step, is far from sufficient important first step, is far from sufficient in the process of systematically improving in the process of systematically improving decision making in the real world of decision making in the real world of business enterprise.business enterprise.Inadequate modeling is just one of the Inadequate modeling is just one of the reasons why decision-makers do not make reasons why decision-makers do not make good decisions.good decisions.
The purpose of this chapter is to help you The purpose of this chapter is to help you understand why improving the quality of understand why improving the quality of modeling alone will not necessarily lead to modeling alone will not necessarily lead to improved real-world decisions.improved real-world decisions.
This chapter will cover critical oversights This chapter will cover critical oversights that users new to the concepts of that users new to the concepts of modeling make in attempting to move modeling make in attempting to move forward to apply those ideas in actual forward to apply those ideas in actual decision-making situations.decision-making situations.The upside and downside potential risks of The upside and downside potential risks of applying modeling concepts will be applying modeling concepts will be discussed discussed so that you will come away with a balanced so that you will come away with a balanced perspective of the pros and cons of perspective of the pros and cons of applying modeling in business practical applying modeling in business practical situations.situations.
WHAT, AFTER ALL, IS A WHAT, AFTER ALL, IS A MODEL?MODEL?
It is difficult to define a model. One It is difficult to define a model. One definition might be:definition might be:
Consider the following evolution of a Consider the following evolution of a model:model:
A model is an abstraction of a business A model is an abstraction of a business situation suitable for spreadsheet analysis situation suitable for spreadsheet analysis to support decision making and provide to support decision making and provide managerial insights.managerial insights.To many managers, a model is exquisitely To many managers, a model is exquisitely
crafted and professionally polished in crafted and professionally polished in appearance, highly intuitive, self-appearance, highly intuitive, self-documenting, easy to use, completely documenting, easy to use, completely validated and generalizable enough to be validated and generalizable enough to be applied in a variety of settings by many applied in a variety of settings by many people.people.
A Prototype ModelA Prototype ModelCompleteCompleteDebuggedDebugged
Runable by Its AuthorRunable by Its AuthorValidated with Test DataValidated with Test DataBelieved to Deliver ValueBelieved to Deliver Value
An Institutionalized ModelAn Institutionalized ModelSustained by the OrganizationSustained by the OrganizationIntegrated into Organization'sIntegrated into Organization's
Decision ProcessesDecision ProcessesCoordinated in Function with Coordinated in Function with
Other Models and SystemsOther Models and SystemsUseable by Other ManagersUseable by Other Managers
Maintainable and Extensible Maintainable and Extensible by Othersby Others
Need Data Supplied and Need Data Supplied and Maintained by OthersMaintained by Others
Effort: 10X-100XEffort: 10X-100XEffort: 1XEffort: 1X
A Modeling ApplicationA Modeling ApplicationUsable by a Client ManagerUsable by a Client Manager
Well DocumentedWell DocumentedHardened to Reject Unusual Hardened to Reject Unusual
Data InputsData InputsExtendable by Author or Client Extendable by Author or Client
Manager Validated with Manager Validated with Real-World DataReal-World Data
Known to Deliver ValueKnown to Deliver Value
Effort: 10XEffort: 10X
An InstitutionalizedAn InstitutionalizedModeling ApplicationModeling Application
Effort: 100X – 1000XEffort: 100X – 1000X
This framework is a variation of one This framework is a variation of one originally proposed by C. West Churchman, originally proposed by C. West Churchman, et. al.et. al.
Modeler,Modeler,Project Project
Manager,Manager,Decision Decision Maker,Maker,ClientClient
Curse ofCurse of
Player SeparationPlayer SeparationClientClient
ModelerModeler
Project ManagerProject Manager
DecisionDecisionMakerMaker
The Separation of Players The Separation of Players CurseCurse
The Curse of Scope CreepThe Curse of Scope Creep
Narrow Modeling ProjectNarrow Modeling ProjectSingle ModelSingle Model
Single ObjectiveSingle ObjectiveFocused ActivityFocused Activity
Few PlayersFew PlayersFew StakeholdersFew Stakeholders
Low EffortLow EffortLow CostLow Cost
Low Development RiskLow Development RiskInformal Coordination & Informal Coordination &
Project ManagementProject ManagementLow Project VisibilityLow Project VisibilityScale Diseconomies in Scale Diseconomies in
Information Systems for Information Systems for ModelModel
Scale Diseconomies in Model Scale Diseconomies in Model & Database Maintenance & Database Maintenance
Deterioration in Model Use Deterioration in Model Use as Early Adopters Move on as Early Adopters Move on
Low Potential Organization-Low Potential Organization-wide Impactwide Impact
Curse ofCurse of
Scope CreepScope Creep
Wide Modeling ProjectWide Modeling ProjectMultiple (Replicated) Multiple (Replicated)
ModelsModelsMultiple ObjectivesMultiple Objectives
Diffused ActivityDiffused ActivityMany PlayersMany Players
Many StakeholdersMany StakeholdersHigh EffortHigh EffortHigh CostHigh Cost
High Development RiskHigh Development RiskFormal Coordination & Formal Coordination & Project ManagementProject ManagementHigh Project VisibilityHigh Project Visibility
Scale economies in Scale economies in Information Systems for Information Systems for
ModelModelScale Economies in model & Scale Economies in model &
Database MaintenanceDatabase MaintenanceSupport for Model Use Support for Model Use Independent of Early Independent of Early
AdoptersAdoptersHigh Potential High Potential
Organizational-wide ImpactOrganizational-wide Impact
Other Frequent Sources Other Frequent Sources of Implementation Failureof Implementation Failure
However, inadequate attention to political However, inadequate attention to political issues that arise from the use of a model is issues that arise from the use of a model is far more prevalent as a source of failure in far more prevalent as a source of failure in modeling.modeling.
Easily addressed issues in modeling failure Easily addressed issues in modeling failure are model logic, model inadequacy, etc.are model logic, model inadequacy, etc.
When a model fails, it is all too common to When a model fails, it is all too common to blame the model when in fact, it was due blame the model when in fact, it was due to inadequacies of the whole process of to inadequacies of the whole process of developing and implementing the model.developing and implementing the model.
Another problem is the potential loss of Another problem is the potential loss of continuity either during the development continuity either during the development of a model itself or later during of a model itself or later during implementation caused by departure of implementation caused by departure of key players, or the key players, or the loss of organizational memory of a loss of organizational memory of a successful model.successful model.A source of difficulty in modeling is the A source of difficulty in modeling is the attempt to develop a modeling application attempt to develop a modeling application before assessing issues of the data before assessing issues of the data availability necessary to support that availability necessary to support that application.application.An important consideration early in the An important consideration early in the model development phase is the matching model development phase is the matching of available data to a possibly less-of available data to a possibly less-adequate model as a adequate model as a way of avoiding implementation problems way of avoiding implementation problems later.later.
An infrastructure must be created that An infrastructure must be created that guarantees that the data and systems will guarantees that the data and systems will be maintained in a way that serves the be maintained in a way that serves the users of users of the model. the model. A more subtle and insidious shortcoming A more subtle and insidious shortcoming of modeling concerns the identification of of modeling concerns the identification of shortcomings at one level of an shortcomings at one level of an organization as being caused by failures or organization as being caused by failures or inadequacies at a higher, often more inadequacies at a higher, often more abstract, level of the organization.abstract, level of the organization.
In this case, the best thing to do is to tune In this case, the best thing to do is to tune the model to work well given other the model to work well given other organizational inadequacies that might be organizational inadequacies that might be addressed more effectively at a later time.addressed more effectively at a later time.