- If someone asked you what type of decision-making model this was, how would you describe it?
- Does this modelling tool (influence diagrams) cater for multiple goals?
- Is your model allowing for uncertainty? if so, how?
- Is this decision probabilistic? Explain why or why not.
- What are heuristics?
- Does this modelling tool allow for Heuristics? Explain
Showing posts with label lesson 2. Show all posts
Showing posts with label lesson 2. Show all posts
Thursday, September 17, 2009
tut
Using the powerpoint slides and the text as resources to help you, answer the following six questions about the model used in activity 1 above:
Model-based management systems
Provide MSS and DSS based upon models of the organisation and its components.
Makes use of software that allows model description and organization with transparent data processing.
Capabilities
Object-oriented model base management system
Database and MIS design model systems
Makes use of software that allows model description and organization with transparent data processing.
Capabilities
- MSS / DSS user has control
- Flexible in design
- Gives feedback
- GUI based
- Reduction of redundancy
- Increase in consistency
- Communication between combined models
- Virtual file
- Virtual relationship
Object-oriented model base management system
- Logical independence
Database and MIS design model systems
- Data diagram, ERD diagrams managed by CASE tools
Simulations page 165
- Imitation of reality
- Allows for experimentation and time compression
- Descriptive, not normative
- Can include complexities, but requires special skills
- Handles unstructured problems
- Optimal solution not guaranteed
- Methodology
- Problem definition
- Construction of model
- Testing and validation
- Design of experiment
- Experimentation
- Evaluation
- Implementation
- Explore problem at hand
- Identify alternative solutions
- Can be object-oriented
- Enhances decision making
- View impacts of decision alternatives
Find-by-search approaches
These are analytical techniques (algorithms) for finding possible outcomes for structured problems
Blind searching
Complete enumeration search
All alternatives explored
Incomplete search
Partial search
Repeated, step-by-step searches
Rule-based, so used for specific situations
Will find a “good enough” solution, but eventually will obtain optimal outcome
Examples of heuristics
Tabu search
Remembers and directs toward higher quality choices
Genetic algorithms
Randomly examines pairs of solutions, chooses best solutions so far and mutates them to find even better solutions
- A general step-by-step search for solutions that eventually obtains (finds) an optimal solution.
Blind searching
Complete enumeration search
All alternatives explored
Incomplete search
Partial search
- Achieves satysficing of particular goal
- May obtain optimal goal
Repeated, step-by-step searches
Rule-based, so used for specific situations
Will find a “good enough” solution, but eventually will obtain optimal outcome
Examples of heuristics
Tabu search
Remembers and directs toward higher quality choices
Genetic algorithms
Randomly examines pairs of solutions, chooses best solutions so far and mutates them to find even better solutions
Sensitivity, what-if, and goal seeking analyses page 159
Sensitivity analysis
- Assesses the impact of change in inputs or parameters on solutions
- Allows for adaptability and flexibility
- Eliminates or reduces variables
- Can be automatic or trial and error
- Assesses solutions based on changes in variables or assumptions
- Backwards approach, starts with goal
- Determines values of inputs needed to achieve goal
- E.g. break-even point determination
Satisfaction of multiple goals page 150
- One decision-making process must try to meet more that one goal.
- Simultaneous goals are often conflicting – achieving one goal makes it harder to achieve the other(s).
- Try to determine a single measure of effectiveness that judges the achievement of all goals.
- Sometimes very difficult to do.
Methods:
- Utility theory.
- Goal programming.
- Linear programming with goals as constraints.
- Point allocation systems.
Mathematical programming page 153
(Not to be confused with computer programming).
Use of defined and verified mathematical processes (algorithms techniques etc.) to determine best outcomes.
Use of defined and verified mathematical processes (algorithms techniques etc.) to determine best outcomes.
- Linear programming
- MinMax techniques
- Game theory
- Tools for solving managerial problems
- Useful when decision-maker must allocate resources amongst competing activities.
- Allocation of resources to achieve optimised specific goals.
Mathematical models
Use mathematical techniques to link decision variables, uncontrollable variables, parameters, and result variables together:
“if the sales this year are $400,000 and they increase by 10% next year, the sales will be $440,000”
Nonquantitative models (qualitative models)
“indicate your happiness with our product on a scale of 1 (very unhappy) to 5 (very happy). …”
Results (outcomes) = f(decision variables, uncontrollable variables, parameters, intermediate results)
- Decision Variables have values that describe choices.
- Uncontrollable Variables have values that are outside the decision-maker’s control.
- Parameters are fixed factors (constants).
- Intermediate Result Variables are intermediate outcomes.
- Result Variables are outcomes dependent on chosen decision variables, uncontrollable variables, parameters and intermediate results.
- Quantitative relationships.
- Variables involved in the model have numerical values.
- Outcomes are numerical.
“if the sales this year are $400,000 and they increase by 10% next year, the sales will be $440,000”
Nonquantitative models (qualitative models)
- Qualitative relationships.
- Variables involved in the model have qualitative values that are represented within the model by numerical values.
- Outcomes are qualitative / numerical.
“indicate your happiness with our product on a scale of 1 (very unhappy) to 5 (very happy). …”
Results (outcomes) = f(decision variables, uncontrollable variables, parameters, intermediate results)
Modeling with decision trees
A graphical representation of a decision.
- A graphical representation of relationships between variables, values and outcomes.
- Each pathway along the ‘branches’ of the tree represents a particular sequence of criteria leading to a particular outcome.
- A multiple criteria approach.
- Can model probabilistic decision-making.
- Cumbersome if there are many alternative outcomes.
Modeling with decision tables
A tabular representation of a decision.
wiki: http://en.wikipedia.org/wiki/Decision_table
Can model multiple criteria decisions
Features include:
Applies principles of certainty, uncertainty, and risk
wiki: http://en.wikipedia.org/wiki/Decision_table
Can model multiple criteria decisions
Features include:
- Decision variables (alternatives)
- Uncontrollable variables
- Result variables
Applies principles of certainty, uncertainty, and risk
Modeling with spreadsheets page 145
see Modelling and Analysis.ppt
Spreadsheets are a computerised mathematical and data management tool.
Spreadsheets include extensive forecatsing, statistical and other modeling and databse management capabilities, functions and routines. As spreadsheet packages eveolved, add-ins were develpoed for structuring and solving specific model classes.
DSS related add-ins
Most popular end user modeling tool, incorporates many power financial statistical, mathematical and other functions.
Spreadsheets can perform model solution tasks such as:
1. Linear programming
Informally, linear programming determines the way to achieve the best outcome (such as maximum profit or lowest cost) in a given mathematical model and given some list of requirements represented as linear equations.
from wiki: http://en.wikipedia.org/wiki/Linear_programming
2. Regression analysis
regression analysis refers to techniques for modeling and analyzing several variables, when the focus is on the relationship between a dependent variable and one or more independent variables.
from wiki: http://en.wikipedia.org/wiki/Regression_analysis
Other features:
http://office.microsoft.com/en-us/excel/CH010004551033.aspx
http://www.informit.com/podcasts/episode.aspx?e=B4C8DE7D-3AC1-45CD-8F4A-1F31714AA61A
http://www.sskkii.gu.se/publications/Documents/html/dynwhatif/
Most spreadsheet packages provide seemless integration beciase they read and write common file structures and easily interface with databases.
Spreadsheets are a computerised mathematical and data management tool.
- An end-user modeling tool.
- Flexible and ‘easy’ to use.
- Supports complex mathematical methods.
- Can be used to implement linear programming techniques.
- Supports complex statistical and regression analysis methods.
- Provides services like what-if analysis, data-base management, automation macros.
- Can be used for static and dynamic modeling.
- risk analysis can be incorporated
Spreadsheets include extensive forecatsing, statistical and other modeling and databse management capabilities, functions and routines. As spreadsheet packages eveolved, add-ins were develpoed for structuring and solving specific model classes.
DSS related add-ins
Most popular end user modeling tool, incorporates many power financial statistical, mathematical and other functions.
Spreadsheets can perform model solution tasks such as:
1. Linear programming
Informally, linear programming determines the way to achieve the best outcome (such as maximum profit or lowest cost) in a given mathematical model and given some list of requirements represented as linear equations.
from wiki: http://en.wikipedia.org/wiki/Linear_programming
2. Regression analysis
regression analysis refers to techniques for modeling and analyzing several variables, when the focus is on the relationship between a dependent variable and one or more independent variables.
from wiki: http://en.wikipedia.org/wiki/Regression_analysis
Other features:
- what-if analysis
http://office.microsoft.com/en-us/excel/CH010004551033.aspx
http://www.informit.com/podcasts/episode.aspx?e=B4C8DE7D-3AC1-45CD-8F4A-1F31714AA61A
http://www.sskkii.gu.se/publications/Documents/html/dynwhatif/
- goal-seeking: Indicating a taget cell,its desired value and changing a cell.
- datamanagement
- programmability (macros)
Most spreadsheet packages provide seemless integration beciase they read and write common file structures and easily interface with databases.
Probabilistic decision-making
Making a decision under risk.
- Several outcomes could occur
- Each outcome has a probability of occurring
- Each outcome has a positive or negative value (payout) associated with it.
- Calculate the value of the outcome for each alternative.
- Calculate the probability of each outcome occurring.
- Calculate the expected value of the outcome for each alternative. (= probability x value)
- Select the alternative with the best expected value.
Reliablility
Certainty:
- Assume complete given knowledge
- All potential outcomes known
- Best outcome determined easily
- Can be very complex
- Incomplete given information
- Several ‘acceptable’ outcomes for each decision
- Probability of occurrence of each outcome unknown
- Assess risk and willingness to take it
- Pessimistic/optimistic approaches
Decision-making
Optimum outcome
Satisfactory outcome – satisfycing.
How sure can we be that a chosen outcome is the correct / best one?
- The absolute best of all possible outcomes.
- Is the most desirable solution to a problem.
- May be difficult or impossible to discover.
- What are the criteria for judging the quality of the outcome?
Satisfactory outcome – satisfycing.
- The best of all determinable outcomes.
- Is an ‘at least satisfactory’ solution to a problem.
- May be easier to discover than the optimum.
- What is a determinable outcome?
- What are the criteria for judging the quality of the outcome?
- When is a satisfactory solution acceptable?
How sure can we be that a chosen outcome is the correct / best one?
Dynamic modeling
Explores a situation as it varies over time.
Time dependant : How may checkouts should be open in the supermarket.
Demands forecasted over time
Dynamic simulation
use, represent, generate trends and patterns over time
- Models changing situations and varying conditions
- Outcomes are time dependent (they may vary with time)
- Explores the impact of trends in the driving variables
Time dependant : How may checkouts should be open in the supermarket.
Demands forecasted over time
Dynamic simulation
use, represent, generate trends and patterns over time
Static modeling page 142
Explores a situation at a single ‘steady’ instant (state).
Assumes the context of the problem is unvarying – or that it can be assumed to be so for simplicity.
Takes a snapshot of the situation - everything occurs at a single interval. For eg a decision to make a certain product is static, quarterly or annual income statement is static.
Static decision making are presumed to repeat with identical conditions.
Dynamic behaviour can be represented by multiple trials of the static model at separate moments.
Assumes the context of the problem is unvarying – or that it can be assumed to be so for simplicity.
Takes a snapshot of the situation - everything occurs at a single interval. For eg a decision to make a certain product is static, quarterly or annual income statement is static.
Static decision making are presumed to repeat with identical conditions.
Dynamic behaviour can be represented by multiple trials of the static model at separate moments.
Categories of models pg 140
Category: Optimisation of problems with few alternatives
- Process & Objective: Find the best solution from a small number of alternatives
- Representative Techniques: Decision Tables & Decision Trees
Category: Optimisation via algorithm
- Process & Objective: Find the best solution from a lareg number of alternatives, using a step by step improvement process.
- Representative Techniques: Linear and other mathematical programming models, network models.
Category: Optimisation via an analytical formula
- Process & Objective: Find the best solution in one step using a formula.
- Representative Techniques: Some inventory models.
Category: Simulation
- Process & Objective: Finding a good enough solution or the best among the alternatives checked using experimentation.
- Representative Techniques: Several types of simulation.
Category: Heuristics
- Process & Objective: Find a good enough solution using rules.
- Representative Techniques: Heuristic programming, expert systems.
Category: Predictive models
- Process & Objective: Predict the future for a given scenario.
- Representative Techniques: Forecasting models, Markov analysis
Category: Other models
- Process & Objective: Solve a what if case using a formula.
- Representative Techniques: Financial modeling, waiting lines.
DSS model types
Generate a possible outcome and present it in a useful way.
Generational strategies
(ways to create a modeled outcome)
(ways to present a modeled outcome)
Generational strategies
(ways to create a modeled outcome)
- Algorithmic: Algorithmic information theory is a subfield of information theory and computer science that concerns itself with the relationship between computation and information.
- Statistical: A statistical model is a set of mathematical equations which describe the behavior of an object of study in terms of random variables and their associated probability distributions. If the model has only one equation it is called a single-equation model, whereas if it has more than one equation, it is known as a multiple-equation model.
- Linear programming: Some industries that use linear programming models include transportation, energy, telecommunications, and manufacturing. It has proved useful in modeling diverse types of problems in planning, routing, scheduling, assignment, and design.
- Simulation
(ways to present a modeled outcome)
- Graphical
- Quantitative
- Qualitative
Modeling and Analysis
Allows for the rapid exploration of several (many) alternative solutions
Each ‘trial’ (run / execution / analysis) of a model explores a particular set of circumstances and generates a likely outcome.
Trials of a model can be repeated (usually with the same outcomes).
A fundamental aspect of DSS methodology
Many classes of models
Some DSSs may incorporate several (multiple) models (hybrid)
Trials of a model can be repeated (usually with the same outcomes)
Each ‘trial’ (run / execution / analysis) of a model explores a particular set of circumstances and generates a likely outcome.
Trials of a model can be repeated (usually with the same outcomes).
A fundamental aspect of DSS methodology
Many classes of models
- specialised techniques for each type
Some DSSs may incorporate several (multiple) models (hybrid)
Trials of a model can be repeated (usually with the same outcomes)
Decision Support Methodology
Match each of the quotations with the corresponding decision-making phase according to Simon's model, note that there may not be a one-to-one mapping. For each phase justify your selection in less than 200 words.
a) "The successful decision maker remains open to the full array of alternative solutions."
b) "Until a decision has degenerated into work and reaches the stage of actual execution, for all intents and purposes there is no decision."
c) "If you have no alternative to fall back on, you begin to drift if the decision doesn't work out."
d) "Organised tracking of progress, results and feedback are non-negotiable elements of any effective action program."
e) "Only by taking the time to investigate what the decision really needs to be about can the decision maker distinguish between the symptom and the ailment."
f) "One does not make unnecessary decisions any more than a good surgeon does unnecessary surgery."
a) "The successful decision maker remains open to the full array of alternative solutions."
b) "Until a decision has degenerated into work and reaches the stage of actual execution, for all intents and purposes there is no decision."
c) "If you have no alternative to fall back on, you begin to drift if the decision doesn't work out."
d) "Organised tracking of progress, results and feedback are non-negotiable elements of any effective action program."
e) "Only by taking the time to investigate what the decision really needs to be about can the decision maker distinguish between the symptom and the ailment."
f) "One does not make unnecessary decisions any more than a good surgeon does unnecessary surgery."
Subscribe to:
Posts (Atom)