Model Driven DSS Chapter 9. What is a Model? A mathematical representation that relates variables For solving a decision problem Convert the decision.

Model Driven DSS

Chapter 9

What is a Model?

• A mathematical representation that relates variables

• For solving a decision problem

• Convert the decision problem into a model

• There can be multiple solutions to a model

• Use math techniques solve the model

Success at workCapabilities

Environmental factors

Opportunities

Help fromManagement

More variables?

Types of models

• Explanatory model– Fitting the data to a model– May be used for forecasting

• Contemplative models– To do what-if type analysis– User Interaction centered

• Algebraic models– Goal seek and optimization

Model driven DSS

• Analytical capabilities; Can answer ‘what-if’ scenarios

• Can be used for deciding which path to take (Goal seek)

• Can be used to determine what inputs will get you the desired output (Solving)

Make or Buy Model based DSS?

• Buy

• Buy and customize

• Very rarely develop from scratch

Software packages

• Statistical modeling

• Forecasting software

• Spreadsheets

• Optimization software

• Financial modeling software

Some popular statistical software packages

Forecasting tools

For more forecasting software visithttp://morris.wharton.upenn.edu/forecast/software.html

Electronic Spreadsheets

Known as DSS generators

For more productshttp://www.dssresources.com/spreadsheets/products.html

Optimization software

MATLAB® 7.2

Financial modeling

Models for accounting and financials

• Break-even analysis – demo at dssresources.com

• Cost-benefit analysis

• Financial budgeting

• Return on investment

• Price determination

Decision Analysis Models

• Muti-attribute utility models– Given a set of alternatives how to choose the best– Consider attributes of alternatives– Try online software at dssresources.com

• Analytical Hierarchical Process– Comparing an alternative to another alternative on

each attribute– Assign a grade between 1 and 9 to record

preferences– Use eigen-values to come up with ranking

Diagrams

• Decision trees– Uses two types of nodes – Choice and chance nodes– Calculate expected payoffs for each branch in the tree

• Influence diagrams– Representation for decision situation– Variables and how they influence one another– Non-cyclical– Types of variables

• Decision (controllable) variable (rectangle)• Chance (uncontrollable) variable ( Circle)• Outcome variable (oval)

– Does not represent temporal events or actions– Develop an influence diagram for some personal decision

Forecasting

• Extrapolation – simple average

• Moving average

• Exponential smoothing (example)

• Regress and econometric models

Optimization models

• What input values will get me the maximal output value?

• Constraints may not be violated

• Linear programming

• Integer programming

• Solver example

Questions?