The Modeling Process Objective Hierarchies Variables and Attributes Influence Diagrams Mathematical Representation Testing and Validation Implementation and use Describe Problem / opportunity Identify Overall Objective Organize Sub-Objectives into a hierarchy Identify Model’s Objective Determine all variables and their attributes Decide on Measurement / Data Collection Graphically depict relationships among variables Distinguish between Decision and outcome variables Determine mathematical relationships among variables Develop mathematical model(s) Evaluate reliability and validity Understand limitations Implement models in DSSs Clarify assumptions, inputs, and outputs
In the context of Values, Mission and Goals Objective Hierarchy Maximize Value to Customers and Stakeholders Fundamental Objectives Profitability Quality Service Reputation Max Revenues Min Costs Means Objectives In the context of Values, Mission and Goals
Variables and Attributes How Measured (Metrics) Variable Data Source
Influence Diagrams A graphical representation of relationships among variables in a given problem situation Informal rules for constructing influence diagrams Identify all variables - decision, intermediate and outcome variables. Connect variables to delineate relationships Identify variables influencing the decision variables - both controllable and uncontrollable. Use consistent representation scheme
Influence Diagrams - an example Market Conditions Room Rate Occupancy Meals Served Room Personnel Room Supplies Room Revenue Operating Cost Meal Cost Meal Revenue Restaurant Margin Room Margin Fixed Cost Profit
Establishing Causality To establish causality (X causes Y), one must show: X precedes Y Most cases easy to demonstrate. A change in either is associated with a change in the other Statistics can establish this. There is no other plausible explanation for the occurrence of Y but X. Impossible in the literal sense
Classification of Models By what the model does Descriptive Prescriptive Optimization By type of problem Forecasting Decision Analysis Constrained Optimization Monte Carlo Simulation
Forecasting Methodologies Qualitative Delphi Methods Quantitative - Non-statistical Using “comparables” Quantitative - Statistical Time-series Regression
Forecasting Models By data type Time series (surrogate variable) Exponential smoothing Moving averages Cross sectional (causal variables) Multiple linear regression