Uploaded by Aka anonymous

Enhancing Decision Making(1)

advertisement
Management Information Systems
IS 601
Spring 2023
Enhancing Decision Making & Business Analytics
Part I:
History & Context for Decision
Making
2
Decision Making
• Three main reasons why investments in information
technology do not always produce positive results
1.
Information quality
• High-quality decisions require high-quality information
2. Management filters
• Managers have selective attention and have variety of
biases that reject information that does not conform to
prior conceptions
3. Organizational inertia and politics
• Strong forces within organizations resist making decisions
calling for major change
Decision Making in the Firm
Decision Making
Herbert Simon
Decision Making Continuum
Bounded rationality: Decision makers face uncertainty due to cognitive
limitations, problem complexity, and time available to acquire information.
Satisficing: to set an aspiration level that one will accept (or settle for) over an
optimal solution. When in doubt, acquire more data!
Mintzberg Managerial Roles
• Information systems can
only assist in some of the
roles played by managers
• Classical model of
management: five
functions
– Planning, organizing,
coordinating, deciding,
and controlling
DSS Components
•
•
•
•
Components
• Database management module
• Model management module
• Dialog module
Database Management:
• Provides data access and the means to select
data for further analysis
Model Base Management:
• Contains statistical, financial, optimization, or
simulation models
Dialogue module:
• Graphical user interface of the DSS which
facilitates changing parameters and
interaction with the program
8
Business Intelligence vs Decision Support Systems
• “Interactive computer-based systems, which help decision
makers utilize data and models to solve unstructured
problems”, Gorry and Scott-Morton, 1971
• “Decision support systems couple the intellectual resources of
individuals with the capabilities of the computer to improve the
quality of decisions”, Keen and Scott-Morton, 1978
• Key differences between Business Intelligence and DSS
–
–
–
–
–
BI is a business strategy whereas DSS is a decision making methodology
BI is a broader term reflecting vendor supported solutions & technologies
DSS is usually associated with more specific functions and algorithms
BI typically utilizes a Data Warehouse whereas DSS utilizes a model base
BI (COTS applications) vs. DSS (Custom development)
Six Elements of Business Intelligence
Business Intelligence and Analytics
• Business intelligence
– Describes how organizations collect, store, clean,
and disseminate information
• Business analytics
– Tools and techniques for analyzing data
– OLAP, statistics, models, data mining
• Most well-known business intelligence vendors
– Oracle, SAP, IBM, Microsoft, and SAS
Business Intelligence and Analytics
• Goal is to deliver accurate real-time information to
decision makers
–
–
–
–
–
–
Production reports
Parameterized reports
Dashboards/scorecards
Ad hoc query & search
Drill down capabilities
Forecasts, scenarios, models
Business Intelligence: Production Reports
BUSINESS FUNCTIONAL AREA
PRODUCTION REPORTS
Sale
Forecast sales; sales team performance; cross-selling;
sales cycle times
Service/call center
Customer satisfaction; service cost; resolution rates;
churn rates
Marketing
Campaign effectiveness; loyalty and attrition; market
basket analysis
Procurement and support
Direct and indirect spending; off-contract purchases;
supplier performance
Supply chain
Backlog; fulfillment status; order cycle time; bill of
materials analysis
Financials
General ledger; accounts receivable and payable; cash
flow; profitability
Human resources
Employee productivity; compensation; workforce
demographics; retention
Business Analytics Models
•
Business analytics (BA) or Business Intelligence (BI): the process of developing actionable
decisions or recommendations for actions based on insights generated from historical data.
Business analytics examines data with a variety of tools and techniques, formulates
descriptive, predictive, and prescriptive models, and communicates these results to
organizational decision makers.
– Descriptive analytics summarize what has happened in the past and allow decision
makers to learn from past behaviors. Examples include reports that provide historical
insights regarding an organization's production, financials, operations, sales, finance,
inventory, and customers.
– Predictive analytics examine recent data in order to detect patterns and predict future
outcomes and trends. Predictive analytics provide estimates about the likelihood of a
future outcome. The purpose of predictive analytics is not to tell decision makers what
will happen in the future. Predictive analytics can only forecast what might happen in
the future, because predictive analytics are based on probabilities.
– Prescriptive analytics go beyond descriptive and predictive models by recommending
one or more courses of action and showing the likely outcome of each decision.
Prescriptive analytics attempt to quantify the effect of future decisions to advise on
14
possible outcomes.
Decision Characteristics
• Structured problems
– Encountered repeatedly
– Known variables
• Unstructured problems
– Often require customized solutions
– May only partially supported quantitative methods
• Semi-structured problems
– Most common
– Involves a combination of standard solutions + human judgment
• Our goal should be reduce uncertainty wherever possible
– Unstructured  structured problems
Decision Characteristics
(Gory and Scott-Morten, 1971)
Balanced Scorecard Method
• The balance scorecard method is an ESS decision support for
senior management
• Help executives focus on important performance information
• Measures outcomes on four dimensions:
–
–
–
–
Financial
Business process
Customer
Learning and growth
• Key performance indicators (KPIs) are metrics which provide
measurement data for each dimension
Balanced Scorecard Method
Part II:
Design Models
19
What is a design model?
• Every DSS has a design model. A model is an abstraction.
• Iconic model: a physical replica of a system, usually on a
different scale from the original
• Analog model: more abstract than an iconic model and is
a symbolic representation of reality (blueprints, maps)
• Mathematical model: Most DSS analyses are performed
numerically with quantitative models.
• Choosing the best model is critically important aspect of the
design phase of IDC.
20
Building a Quantitative Model
Uncontrollable
variables
Decision
Variables
Mathematical
Relationships
Decision Variables
•
•
•
Describe alternative courses of action
The decision maker controls them
Independent variables
Result Variables
•
•
Reflect the level of effectiveness of the system
Dependent variables
Results of Decisions are Determined by the
•
•
•
Decision
Relationships among Variables
Uncontrollable Factors
Result
Variables
Data Preparation
• Data preparation (or cleaning) is an important first step prior
to quantitative analysis for a design model.
• A scientist systematically corrects data abnormalities or errors
and determines the types of statistical tests to be performed.
• This includes assessing the quality of data sets or excluding a
particular data set from the analysis.
• Therefore, data cleaning results in a reduced “Data Analysis
Set” which can be used for further testing
• Let’s look at an example…
25
2
36
Methods for Evaluation of Alternatives
• Once we understand the problem, how do we choose from
potential alternative courses of action generated from design?
• There are multiple and often conflicting goals
• Evaluation of the search space often leads to an optimal solution
• Methods for evaluation
–
–
–
–
–
Multi-Dimensional Analysis (e.g. OLAP)
Group Decision Support Systems (GDSS)
Sensitivity Analysis
Heuristic Programming
There are many more!
Sensitivity Analysis
• Refers to the study of how the variation (uncertainty) in the
output of a mathematical model can be affected, qualitatively
or quantitatively, by different sources of variation in the input
of a model
– Change inputs / parameters
– Look at the results
• Types
– Automatic
• Linear Programming
– Manual
• What - if analysis
• Change input data and re-solve the problem
• Better and better solutions can be discovered
Heuristic Programming
• Optimization via algorithm
• Heuristics can be
– Qualitative
– Quantitative
• Heuristic programming reduces the
problem / search space
• Identifies solutions quickly
• Provides feasible solutions to
complex problems
• The Traveling Salesman Problem
(TSP) is an example of heuristic
programming via optimization
algorithm
Traveling Salesman Problem
• Most intensively studied programs in mathematics
– The Traveling Salesman Problem (TSP) and the
various algorithms to solve it are based upon
graph theory which is the study modeling
pairwise relations between objects (vectors)
– A graph or is made up of vertices or points which
are connected by edges.
• Traveling salesman must travel between cities at the
lowest possible cost
• Goal: TSP computes
– Shortest route that visits every city and returns to
the starting point.
Conclusions
• Decision making is a complex organizational and human centered activity.
Have a well thought out methodology is critical to arrive at the optimal
solution. In conjunction with managerial decision making, Simon’s
Intelligence, Design, Choice & Implementation Model provides a basis for
solving research problems
• The balanced scorecard is a framework for operationalizing a firm’s
strategic plan by focusing on measurable outcomes in the areas of financial,
business process, customer orientation, learning and growth.
• Selection of a business intelligence vendor supported solution (descriptive,
prescriptive or predictive models) may require additional framing, problem
analysis, and data-sourcing activities (when in doubt – acquire more data!)
• Various methods for the evaluation of alternative courses of action rely
upon quantitative design models and inevitably lead to optimization (or
even sub-optimization) problems for business intelligence
Download