Decision Making Contents Basics Decision Decision Types Decisions
Decision Making
Contents Basics: Decision, Decision Types, Decisions Makers The Decision Making Process Structured Decision Making Representing Structured Problems
The Notion of a Decision A decision is a commitment to an action, course of action or strategy Decision Commitment Attend class today Action of attending class Register for a course Course of action • Attend all classes • Be punctual • Do all non-contact work • Do all exams, assignments etc. Become the best badminton player Strategy • Improve attacking game Commitment implies decision triggers the associated action/course of action the triggered action is done to its completion
A Good Decision A good decision results in the attainment of the objective(s) that gave rise to the need for the decision within the boundaries and constraints of the problem context Decision Objective(s) Attend class today Keep myself occupied Meet attendance norm Better educate myself Register for a course Meet minimum registration credits Meet core requirement for obtaining degree Strengthen my chosen specialization Limitations of the problem context: time, resources A good decision solves the problem at hand without creating a new one
Constraints and Boundaries Personal Needs and Emotions • Feelings • Health • Rewards • Anxieties • Intellectual limitations Environment/Economic Factors • Resources • Regulations • Ethics/Morality • Market demand • Stakeholder interests • New technology Organizational Factors • Policies and business rules • Group conformity • Communication practice Decision Maker Problem context itself • Time required • Perception of importance of decision maker • Motivation to reach decision • Skill inventory to reach decision
Types of Decisions Decision Negotiation Based Extent to which objectives and procedures involved are clear Activity based Nature of activities with which the decision is associated Strategy based The strategy to be adopted in making the decision
Types of Decisions Class Decision Explanation Negotiation Routine Clarity on goals, procedures. technologies Creative Uncertain goals, unknown procedures Negotiated Conflicts in goals/approaches. Need consensus Entrepreneurial High uncertainty. Proactive for near-term growth Adaptive High Uncertainty. React to near term issues Planning High risk. Proactive and reactive for long term growth and efficiency Computational High certainty of outcomes and cause/effect relationships. Strong preferences for possible outcomes Judgmental Highly uncertain outcomes and cause-effect. Strong preferences for possible outcomes Compromise High certainty for outcomes and cause-effect but weak preferences for outcomes Inspirational High uncertainty and weak preferences Activity based Strategy based
Types of Decisions: Another View Abstracting from the classifications, we get Routine, recurring with high certainty Non-routine, non-recurring with high uncertainty Type Nature Properties of Decision making Process Structured Routine, repetitive, high certainty • Easily determined objectives with no conflicts • Clearly defined actions, courses of action, strategies • Clear evaluation criteria • Ascertainable outcomes Ill structured/ Unstructured Non-routine, non- • Objectives of situation conflict recurring with • Actions/courses of action/strategies high uncertainty difficult to isolate • Outcome of alternative is highly uncertain Semi-structured Intermediate • Partly structured and partly unstructured
Exercise: Type of Decision How much salary to pay to an employee at the end of the month? • What are the components of salary? Are these known? Additive components are Basic, AGP, HRA, DA, CCA, Transport, ……. Subtractive components are loan, tax, lease rent, utility charges, Sometimes a wild card is introduced: Special pay, Special deductions • Are the values of components known? Tax to be deducted at source in the year is not known till the last month • Are the contributions of the components known? DA = 40% of (Basic + AGP) HRA = 25% of (Basic +AGP + DA) • Is the procedure of computation known? Yes. Add all additive components and subtract all subtractive components How to account for days not worked? Strikes, disruptions in city/workplace Different types of employees: contractual, regular, ad-hoc, casual
Exercise (Contd. ) Other Issues Can salary be paid? Not enough bank balance Signing Authority is not available Company is closing down: Kingfisher Airlines employees!! Suspended employees New pay fixing formula Court orders: Pay/Do not pay/keep on hold Last salaries: resignations, dismissals, superannuation, deputations, external assignments Inter-departmental relationships Leave record is kept in HR department; Salary is prepared by Accounts Department DA, HRA percentages declared by HR Employees being paid from several account heads Paid from different head from where they work
Decision Makers: Who make Decisions Individual decision maker: works alone to reach a decision independently analyses information and finally generates the decision Multiple Decision makers: have a stake in a particular decision outcome multiple individuals interact to reach a decision Distributed authority to take a decision Reach decision by agreement for a common commitment to course of action Informal communication channels are used Group Decision Makers: stake in a particular decision outcome membership in formal group structure: committees members have equal say in decision formation Team Decision Makers Authority to take decision rests in an individual Team members empowered by the individual assist in decision making Organizational: at highest level of the organization cross cutting decisions supported by the full enterprise for implementation
An Organization Model
The Decision Making Process
The Decision Making Process: Simon’s Model Selects analyzed alternative Decision maker looks for information suggesting occurrence of problem Reality of Situation Choice Design Intelligence Model validation Solution testing success Forming and analyzing alternative solutions, identifying criteria and problem solving strategy Implementation Outcome Failure
Rationality and Bounded Rationality Economic models are based on rationality: Supply and demand, market equilibrium Rational Decision-making: Humans look for only the optimal solution from among all possible solutions. Optimal is rational. Maximization of profits for business Maximization of utility for people Optimizing Strategies • • • Linear programming Goal programming Simple queuing models Investment models Inventory models Transportation models
Rationality and Bounded Rationality Managerial Decision-making are not rational in this sense qualitative and not amenable to quantitative methods build new Institute cost of land etc. are qualitative quality of life of faculty/staff are qualitative search for all possible solutions is not viable comparison and analysis is beyond our cognitive limits Satisfycing Strategies • • • Simulation Forecasting What-if analysis Markov analysis Complex queuing models Environment Impact Analysis
Rationality and Bounded Rationality Simon proposed bounded rationality decision maker develops a model of the likely solution searches problem space for an alternative that fits the model and selects it The optimal solution may not be the one chosen The selected solution “”satisfices” the problem The decision maker is unlikely to spend the energy and cost to gather all information regarding a decision Even if all information is available, the decision maker is unlikely to use all of it due to cognitive overload Decision maker has a preconceived notion of the likely solution Decision makers work on the basis of heuristics when searching for a solution
Bounded Rationality: Example Radha complains of headache while attending class. What should we do? Perceived condition: Radha has a headache Desired condition: get rid of the headache Bounded rationality: give medication to remedy headache. Paracetamol Result: Success, desired condition achieved Next day repeats, third day repeats Bounded rationality is treating symptoms rather than the disease
Bounded Rationality: Exercise Review shows that the price of an article has risen very high Is this a problem? The sales of the article have not been affected Do we have a problem? Sales have remained constant since a manager has offered steep discounts to maintain market share Is there a problem?
Structured Decision Making
Steps in supporting Structured Decision Making Define Problem 1 2 Define Issues, Objectives & Evaluation Criteria 3 Develop Alternatives Estimate Consequences 4 5 Make Trade-Offs and Select 6 Iterate as required Implement and Monitor
Key insights: 1) Separate facts and values 2) Measure objectives in terms that make sense 3) Create alternatives responsive to objectives 4) Simplify decision elements when possible 5) Don’t ignore uncertainty or data gaps 6) Stay flexible and incorporate what is learned 7) Involve stakeholders in both analysis and dialogue 8) Don’t lose track of common sense 14
Structured Decision Making methods spend more time at the start of a decision making process, working with stakeholders to bound the problem and to develop a comprehensive set of concerns Success is measured by the ability of management alternatives to meet peoples’ fundamental concerns and to be implemented in a timely manner 15
Step 1: DEFINE THE PROBLEM What is the problem, or the set of problems, that should addressed? be Beware of Bounded Rationality!! Who is involved (stakeholders)? What concerns are included? -- social, economic, environmental & health -- benefits, costs, and risks What options are possible? What are goals of decision makers? 16
STEP 2: Identify Issues Objectives and Measures Issues : Stakeholder concerns that are to be addressed The broad desires to be considered Problem: Where to go on a holiday? Issues Must be affordable Destination not too far away Availability of medical facility Destination must have good weather 18 Issues are “fuzzy”, subject to different interpretation
STEP 2: Identify Issues Objectives and Measures Objectives: what you want to achieve give body to Issues Issue: Improve financial health To define an objective state � What is valued � Profit � An adjective to describe desired direction of change (more vs less) � Higher the better, Maximize Objectives can be quantitative or qualitative, “hard” or “soft” 18 Objectives should be independent: knowledge of an objective has no bearing on achievement of the other
STEP 2: Identify Issues Objectives and Measures Objectives need to be attributes/measures/criteria • Objective: Maximize Profit • Attribute: Rupees “operationalized” by identifying appropriate Attributes provide a scale for measuring the degree to which its objective is met. Attributes should be Comprehensive: knowing the value of an attribute, we get a clear understanding of the extent that the objective is achieved. Measurable: we can assign a value AND assess the decision maker’s preferences for the different values of the attribute. Relevant, and not subject to other extraneous considerations.
Forming Objectives and Criteria Purchase Flight ticket O Typical ● ● ● ● ● issues: I don’t want to spend much money I don’t want hidden fees I don’t want to spend an extra day in Mumbai I want a direct flight I want easy check-ins I want decent leg room I want an aisle seat I want friendly service I am concerned about all the airline crashes recently I am not comfortable flying with a new airline
Forming Objectives and Criteria Issues OI don’t want to spend much money O I don’t want hidden fees OI Objectives € Minimize Cost $ Total € Minimize Travel Hours € Maximize Comfort Scale (5 = best, 0 = Worst) € Maximize Safety # Accidents / 1 million take-offs (5 yr avg) want a direct flight I don’t want to spend an extra day in Mumbai want decent leg room O I want an aisle seat O I want friendly service O I want easy check-ins Time OI OI Evaluation Criteria am concerned about all the airline crashes recently O I am not comfortable flying with a new airline
Forming Objectives and Criteria Objective Indicator Units Preferred Direction Minimize Cost $ Lower is better Minimize Travel Time Maximize Comfort Hours Lower is better (5 = best, 0 = worst) Higher is better Maximize Safety # Accidents / 1 million takeoffs (5 yr ave) Lower is Better A B Air A Air B C Air C
Evaluating Objectives and Criteria Objective Indicator Units Preferred Direction A B Air A Air B Air C C Minimize Cost Rs. Lower is better 8, 000 7, 500 3000 Minimize Travel Time Maximize Comfort Hours Lower is better 8 -9 13 -15 12 -64 (5 = best, 0 = worst) Higher is better 4 4 3 Maximize Safety # Accidents / 1 million takeoffs (5 yr avg) Lower is Better 3. 8 2. 6 ? don’t know
Structure and Types of Objectives End or Fundamental objectives Why the decision maker cares about the decision The basis on which available alternatives should be evaluated Can be decomposed hierarchically into a tree: no overlap between different branches of the tree Means objectives Help to achieve fundamental objectives Important only because of its implications for some other objective Organized into a network of objectives
Structure and Types of Objectives • A means objective is one way to achieve another objective. • A means objective network is a graphical tool to connect objectives, depicting the means-ends relationships. – Each objective is written in the network and then the objectives are connected by arrows. – An arrow from objective A to objective B indicates that objective A is a means to achieving objective B. • • • I want (objective A) – Why do I want (objective A)? Answer – Because then I will get (objective B) When do we reach fundamental objectives? – The control of consequences criteria tells us A B
Example: End Objective Maximize Safety Minimize Loss of Life Minimize Serious Injuries Minimize Minor Injuries Lecture 03
Example: Means Objective Network Maximize safety Maximize use of vehicle-safety features Motivate purchase of safety features on vehicles Require safety features Educate public about safety Minimize accidents Maximize driving quality Maintain vehicles properly Enforce traffic laws Have reasonable traffic laws Lecture 03 Minimize driving under influence of alcohol
Representing Structured Decision Problems 1. Influence diagrams (decision-focused) 2. Analyzing and communicating uncertainty 3. Improving judgments of experts 4. Decision trees 5. Trade-off analysis tools
Influence Diagrams Also called relevance diagram, decision network
Definition An ID is a directed acyclic graph with 3 kinds of nodes and 3 kinds of edges Decision node: decision to be made Which flight? Uncertainty node: each uncertainty Weather condition Deterministic node: special uncertainty whose outcome is deterministically known if that of another uncertainty is known Visibility is known once weather condition is known Value node: outcome/value obtained by a decision satisfaction Profit
Definition (contd. ) Edges Functional edges: end in value node or decision node value is function of the node at the source decision is dependent on another decision Visibility Satisfaction is a utility function of Visibility and flight satisfaction is quantified if visibility and selected flight known Which flight? satisfaction
Definition(contd. ) Conditional edges ending in uncertainty or deterministic node Uncertainty in deterministic node is deterministically conditioned by the node at the source ( to be considered later) Uncertainty of sink node is probabilistically conditioned on uncertainty of source node Weather forecast Visibility Weather condition
Definition(contd. ) Informational edge ending in a decision a node the decision is made with knowledge of the outcome of the source node the decision is made with knowledge of the weather forecast Which flight?
An Influence Diagram Weather forecast Conditional edges Visibility Weather condition Information edge Functional edges Which flight? satisfaction
Objectives Hierarchy in Influence Diagram An objective may be decomposed into its simpler objectives till atomic objectives are reached Success NET exam? Get scholarship. Become lecturer Career progression
Multiple Objectives A decision may address more than one objective Which battle tank ? Fire power Customer acceptance Battle worthiness Cost
Deterministic Nodes Uncertainty in deterministic node is deterministically conditioned by the source node Fixed Cost Units Sold Cost Revenue Product Price? New product? Profit Variable Cost
Trade-offs • Most important decisions involve multiple objectives • Multiple-objective decisions involve a trade-off – Accept less achievement in terms of some objectives in order to achieve more on other objectives. – But how much less would you accept to achieve how much more?
Trade-offs in Objective Hierarchy
What does an Influence Diagram Provide? Decision Node: available choices/alternatives Outcomes/Objectives: Therefore variables and their measures Links: functional or informational Uncertainties in information Convert the Influence Diagram into a working data warehouse Influence Diagram • Build objective hierarchies • Determine variables and measures • Identify detailed and aggregate information • Identify aggregation procedures Requirements Engineering Process Data warehouse Requirements
Decision Support Systems
Definition of DSS Gory and Scott 1989 DSS: a computer based support system for management decision makers who deal with semi-structured problems. What can a DSS do? Support structured decision making and provide help in unstructured situations • Store and provide information • Provide Intelligence: Compute if there is a problem o determine gap between expected sales and real sales at time t. • Support decision making: Process information o calculate salary to be paid, check against bank balance o process information to suggest course of action • Support decision making: exploring o trend analysis/forecasting o what-if analysis
- Slides: 50