MBA elective - Models for Strategic Planning - Session 1
Models for Strategic Planning Session 1 contents About the Course The Modeling Approach Introduction to Optimization Models
© 2009 Ph. Delquié
1
Models vs. Managerial Judgment Judgment without Models:
you’re limited
Models without Judgment:
you’re dangerous
Models don’t take decisions. People do.
2
MSP Course Material Lecture slides, Solutions and other files distributed via website
Software MS Excel with “Add-ins” • Solver (Part I) • TreePlan (Part II) • Crystal Ball (Part III)
Recommended text The Art of Modeling with Spreadsheets, Powell and Baker, Wiley, 2007 (includes software)
3
Synopsis of Business Analytics
DATA
Prediction Models Help you extract information from data
INFORMATION, KNOWLEDGE
Decision Models Help you exploit information to make better decisions
ACTION
4
Course Contents
Optimization Models extract maximum value from resources
Dynamic Decision Models manage, and profit from, uncertainty
Value Models ensure consistent, rigorous valuations
Individually, suited to different situations
Together, tackle a wide array of business issues, tactical and strategic
Simulation Models evaluate and manage complex risks
5
The Building Blocks of Decision Models
Uncertainties Objectives Outcomes Decisions
Constraints
6
The Modeling Approach – Skills you will learn 1. Framing the issues – Conceptual How should I think about this problem? Am I working on the right problem? Model = management’s best current understanding of the issue ⇒ Not “truth”, not free from judgment, 6
2. Designing – Technical How do I capture this in a model structure? Capture just enough detail to support decision making ⇒ Find a balance between practicality and accuracy
3. Generating insights – Analytical How do I exploit the model to go beyond what I already know? ⇒ Generate solutions; discover alternatives, trade-offs, boundaries 7
The Modeling Approach – Model vs. Reality Model differs from reality - Structurally:
the equations do not correspond precisely to the actual situation
- Parametrically: we are unable to determine all coefficients precisely
⇒ Solution to the model is not an exact solution to the real problem
GIGO principle
“Garbage In, Garbage Out”
The quality of recommendations from a model is as good as the quality of input data.
⇒ Need to carry out Sensitivity Analysis
8
Issues in Building and Using Models
• Lack of standards in business modeling • Bugs can have a substantial impact on business (see e.g. www.EuSpRIG.org)
• Ethics in Decision Modeling? - Frame manipulation - Hidden assumptions - Misrepresentation of stakeholders’ preferences - Outcome manipulation
9
Part I. BUILDING OPTIMIZATION MODELS Purpose: Find how to best use the resources available (e.g., money, labor, time) Decisions
so as to
Achieve a best level of something you care about (e.g. profit, cost, risk) Objectives
Helps you extract maximum value from resources, activities, and portfolios Allows you to explore vast, complex combinations of possibilities Enables you to develop insights into the key trade-offs inherent in your business activities 10
Guiding principle for building an Optimization Model Describe the whole situation in terms of three elements: 1. Decision Variables 2. An Objective 3. Constraints A surprisingly wide array of management problems can be thought of in those terms Let’s review these three model elements in turn6
11
Building an Optimization Model (1) 1. Decision variables • Represent parameters under management’s control, to be decided by management • Should be defined so as to describe all possible alternative decisions Examples:
- whether to support an R&D project or not - number of salespeople to hire - production level in a given period - amount to buy from a given supplier - how much to bid
Together, the values of the decision variables define a policy or plan of action.
12
Building an Optimization Model (2) 2. Objective function • Defines the goal of the problem, the target to be optimized • Provides a criterion to compare alternative solutions • Expressed as a function of the decision variables Examples:
Profit Cost Risk Market Share
→ → → →
Maximize Minimize Minimize Maximize
Objective function = an outcome variable can be controlled through decision variables, not directly 13
Building an Optimization Model (3) 3. Constraints • Describe what’s actually feasible6 - technically quantity produced cannot exceed production capacity - legally employees may not work more than 8 hours per shift - logically all parts of the budget must add up to 100%
• Must be expressed as a function of the decision variables
14
In summary... An optimization model is defined by - decision variables X1, X2, . . . , Xn whose values we want to decide
- an objective function f (X1, X2, . . . , Xn) which describes the quantity to be optimized, i.e. the objective of the problem
- constraints gj(X1, X2, . . . , Xn) ≤ bj j = 1, . . ., m which reflect the economic, legal and technical realities under which we must operate
Issue: what if there are multiple objectives? 15
Optimization Models: a worked-out example Mathematical Model Formulation: • Decision variables X1 = number of MP3 Players to produce X2 = number of CD Players to produce X3 = number of Radio Players to produce
• Objective function Maximize: Profit = $75·X1 + $50·X2 + $40·X3
• Constraints 50,000 ≤ X1 ≤ 150,000 50,000 ≤ X2 ≤ 100,000 50,000 ≤ X3 ≤ 90,000 3·X1 + 2·X2 + 1·X3 ≤ 400,000 hrs
16
A spreadsheet implementation of the model:
17
Optimization with Excel’s Solver Solver is found in the Tools menu (if not, go to Add-Ins6) • Solver Parameters dialog box Set Cell = the Objective Function (select Max or Min) Changing Cells = the Decision Variables Constraints = the Constraints
• Solver Options dialog box To specify that all decision variables must be ≥ 0, check the Assume Non-Negative box !
18
To do for next session< Review Solution Set 1 to be posted on website shortly
Prepare Exercise Set 2 Form a workgroup of 5 or 4 members send email to
[email protected]
19