Transcript Lecture 5 ( August 31, 2002)

Lecture 5
( February 15, 2003)
Decisions and models
Case Analysis
Package delivery industry: Federal Express
Business Decisions
• In lecture 2, we talk about the traditional role of
management and their job activities
• Let us review the decision process
–
–
–
–
–
Collecting data
Identifying problems
Making choices
Persuading others to accept a decision
Implementing the solution
• Using information systems to make better decisions
–
–
–
–
–
Improve access to data
Evaluating variables and choosing alternatives
Build models for complex cases
Evaluate and organize models
Display output
Models
• A model is a simplified, abstract representation of a real-world
system. It can be a mathematical expression, graph or even
subjective description. In actual practice, a model could be very
complex
• Models help managers visualize physical objects and business
processes
• Models are important to analyze problems and make decisions
• Output of models can be exact or subject to interpretation
• Businesses use models of the past (to improve processes) to help
with the present (to evaluate choices) and to guide the future (to
forecasting alternatives)
• Information systems help build models. Information systems can be
models themselves. Enterprise information systems are models of
the entire business.
Types of models
• Physical:
– Miniature building an architect might build
– Replicates of intended products
– Tools available, e.g. computer-aided design (CAD)
• Process:
– Process models are symbolic or descriptive, such as diagrams or
graphs.
– We often use models and pictures to simulate objects and
mathematical relationships to represent processes
• Business:
– Business models help describe businesses and business
decisions.
– Dividing a company into functional departments is a business
model.
– Reengineering is a business modeling technique that has gained
momentum as corporations have tried to become more efficient
Application of models
• Optimization:
– Using mathematics or other analytical tools that
evaluate different alternatives while choosing the best
decision.
• Prediction:
– Based on an historical approach and develops a
projection of what the system should look like
• Simulation:
– The modeling technique applies a model to the effects
of changes and situations on the item being studied.
These models are particularly useful for experiments
that would not be safe to perform in real life.
Model Building
• Assumptions: since models are built to simplify a real-life
situation, assumptions must be built into the model that
are reasonable, accurate and well-communicated.
• Input/Output variables: choosing the correct variable is
very important. The selected input variables must be
correlated to the output variables identified for analysis.
• Process/Equations: it is important to identify and
understand the processes that are represented in the
equation and calculations at the process model
• Software, if used, can be generic or pre-programmed
and specific. Parameterized control give users the
flexibility but too many parameters also confuse users.
Limitations and Errors
• Models can be expensive to build, both in terms
of time and dollars. Budgetary and time-frame
considerations are important in the evaluation
process.
• As models are simulations of real life situations,
there are more opportunities for errors in the
assumptions built into the model.
• Main types of error:
– Mistakes in input data
– Errors in equation used in the model
– Flaws in the display or interpretation of the results
Decision Support Systems (DSS)
• DSS provides support through data collection, analysis
of models and the presentation of output. DSSs consist
primarily of a database or data warehouse, modeling
tools and presentation software.
• DSS outline, categorize and weigh factors that need to
be combined to develop a decision. They help users
make tactical decisions through the use of data, models
and presentations to map and solve general problems.
• Data collection is typically performed by the transactionprocessing systems. If the transaction system is not
working properly, the DSS will not work either. The
fundamental difference between a transaction
processing system and a DSS is the support for creating
and evaluating models.
More on DSSs
• DSS requires more than just data. The quality and availability of
data are so important to decision makers. How data are retrieve,
shared and transferred play a vital role. Under legacy systems, it is
difficult to get access to the transaction data. These systems were
designed to collect and store huge amounts of data very quickly.
The systems often cannot handle the additional load of providing
searches and aggregations needed for making decisions efficiently.
In addition, different systems may have incompatible data formats.
This is where data warehousing comes into play.
• DSS typically have features to query data, analyze and store
models.
• Output is another important features of DSSs. If a DSS cannot
produce output in a format that is easy to understand, then it will not
be useful. Graphs and charts are typical outputs people find it easier
to comprehend.
• DSS can be classified broadly into generic and preprogrammed.
Generic and Preprogrammed DSSs
Generic software can be applied to any situation or business but one
need to build the models himself. Spreadsheets can be
considered as a generic tool. Statistical capabilities such as
regression, mathematical formula and predefined functions are
available and hence suitable for finance and accounting
problems.
Preprogrammed software services specific problems. However, some
situations may occur in many different businesses, there are
preprogrammed software that can be tailored for use by just
altering some input parameters or variables or even equations.
Evaluation and trial are important in this case if it can be applied
correctly to your specific problems. Examples are stock market
models and economic models.
Expert Systems (ES)
• An expert system is a software program that structures a problem to
enable the computer to methodologically step down the identified
issues to a resolution.
• Usually built on an expert system shell or platform, the application
incorporates forward or backward chaining.
• Forward chaining begins with a problem and steps through its
resolutions by making decisions about specific steps that could
logically be followed. Software that provides the template for wills,
contracts and trusts are examples of forward chaining.
• Backward chaining begins with the solution or present situation and
steps backward through the process to determine how the issue
started. Medical diagnosis and legal analysis software are examples
of backward chaining.
• Implementation of expert systems require an expert engine, a rule
base and an expert to moderate rules that are not accurate and
make new rules to continue to be applied to new situations.
Artificial Intelligence (AI)
• Artificial intelligence is the effort to “teach” the computers
to “think” using logical steps, if/then analysis, and
specialized software. The power of the computer is
implemented in its ability to “learn” by listing and
searching all the possible solutions that are available, or
to use pattern matching to determine the sequence in
which particular functions are followed.
• In this way, machines can “think” like human beings.
• Although there are limited business applications to much
of this current research, there are two main reasons for
staying abreast of the capabilities. First, anything that
makes the computer easier to use will make it more
useful, and these techniques continue to improve.
Second, one need to understand the current limitations to
avoid costly mistakes.
Areas of AI Applications
• ESs may be considered the most
commercially prominent area of AI
applications. Often, it is separated from
the general term of AI systems.
• Other areas such as robotics, computer
vision, natural language processing ,
speech recognition and machine learning
are more prominent in research areas but
yet limited in practice.
Differences between DSS, ES and AI
• This is best illustrated by an example.
• Take an inventory system which determines when an
item should be re-ordered and the method used to
accomplish it:
– A DSS would collect sales and cost data and automatically apply
the chosen inventory searching method to monitor sales to send
messages to suppliers when a re-order was needed.
– An expert system would help managers decide which inventory
re-ordering method to use when asked for each product or store.
– An artificial intelligence system would determine the rules the
expert system needs to make a decision. This would enable the
manager to switch back and forth between inventory re-ordering
methods whenever applicable.
Problems with artificial intelligence
Humans are significantly superior to computers in six
areas:
1) Pattern recognition
2) Performing multiple tasks at the same time
3) Movement
4) Speech recognition
5) Vision
6) Language comprehension
Of late, computer systems and tools have made great
strides to close gap between humans and computers
in some of these areas, such as speech recognition,
there are still significant gaps in performance in others.
combining human and computer strengths
However, computers have their own strengths over
human beings in other areas:
1) Speed
2) Accuracy
3) Memory
4) Storage capacity
5) Inter-connectivity
Information systems are all about how we make
use of these strengths to our advantage.
Federal Express
• Founded by Fred Smith, an undergraduate at Yale
University in 1971
• The idea was to provide overnight delivery of small, highvalue items.
• World’s largest express transportation company by now.
• Delivers 5 million items each business day to 211
countries worldwide
• 44,000 posting stations worldwide
• Processed 110 million electronic transactions a day
• Employs 120,000 people worldwide
• Operates 644 aircraft and more than 67,200 vehicles in
its integrated system
• Technology has been integral to FedEx’s growth and
success. FedEx pioneered the first automated customer
services center
Using technology
• Customers are willing to pay a premium for
door-to-door service. FedEx had to pick,
transport, and deliver packages to and from the
most lucrative cities as efficiently as possible.
• FedEx developed a three-model management
planning system to meet with the increased
pressures. It used these models to make both
ongoing operational decisions and crucial
strategic decisions.
Three-model management
planning system
• Origin-destination flow model: This model used
an improved origin-destination flow approach to
determine the what, when, and where of package
volumes from and to actual and potential cities in the
system.
• FLY model: This model produced schedules and
determined resource requirements for selected cities.
Using actual past volumes to review performance, this
model tested other options and recalibrated its
coefficients.
• Financial planning model: This model examined the
overall economic and financial implications of alternative
route structures and flying schedules.
More technology
• To have optimal margins, very high load factors is vital. FedEx
reconfigured its route structure every month. It implemented
schedule changes within a few days.
• To improve customer service, FedEx developed a reservation
system. Customer calls were centralized at call centers. Call
centers sent requests to dispatch centers in the cities served.
• FedEx established two hubs that ran in parallel to cope with growth.
However, the debate of single-hub and multi-hub systems went on.
• In 1979, the massive, highly automated SuperHub system was
implemented.
• The SuperHub and centralized call centers moved FedEx from a
decentralized to a centralized structure with strict standards and few
redundancies.
• From SuperHub to overlay hub: In an overlay hub system each
pickup station makes a decision to send a package for re-distribution
within the region by sending it to the regional hub or to send the
package to the SuperHub.
and onto the net
• As volume growth accelerated rapidly, FedEx enhanced
its system into a mainframe-based order and dispatching
system called COSMOS (Customer, Operations, Service,
Management Operating System)
• COSMOS keep track of every package through the entire
FedEx system by using bar codes and scanning
• The growth of PCs had lead FedEx to use electronic
commerce over the internet to capture customers and
reduce costs.
• Fedex.com started in 1995 and expanded progressively.
By 1997, the hits to the site already reach 280,000 per
day.
• FedEx’s extranet, a secure communication and data
exchange between internal and external users was
established integrating the elements of both the intranet
and internet.
other technology stories
• Significant investment in technology, including system of
electronic data lines and hardware necessary to send
and receive data across those lines, were made to
provide a proprietary fax service called Zapmail.
• However, the rapid development of a single fax
technology standard and the nearly instant price decline
that made fax machine ubiquitous. Even small
companies and individuals can afford to have their own
machines which make FedEx fax service redundant.
• In 1996,FedEx terminated its development of a version
of its FedEx Ship Software for Lotus Notes. As FedEx
Ship for Notes required users to have the groupware
Lotus Notes which become outdated.
services at customer sites
• IntraNetShip: a workgroup-enabled extranet application that links to
customer intranets and automates packet tracking and authorization
processes. IntraNetShip runs on customer servers and interfaces
with fedex.com. The application centralizes policy management at
user sites and standardizes authorization procedures. It represents
an expansion of FedEX’s internet strategy, from web-based package
tracking and other services to a model based on server applications
at customer sites.
• FedEx PowerShip PassPort System: gives a customer corporation
the secure means to efficiently meet the varied requirements of its
busy shipping department. Designed to be integrated into
customer’s computer network, this customized system delivers the
power of FedEx's global IT network to customer’s shipping
department.
other fedex.com services
• FedEx Ship Manager at fedex.com: FedEx Ship Manager at
fedex.com gives customer the ability to ship and track from any
computer with Internet access, without the need of any additional
software.
• FedEx's eBusiness Tools are a series of web and Java-based
systems that allow one to seamlessly manage inventory, on-line
shipping, order processing, return management and other logistics
operations.
• FedEx's eShipping Tools make use of the latest information
technology to enable one to efficiently fulfill shipping requirements
from desktops, saving both time and money.
• FedEx is beta testing new software that enable corporate customers
to take a virtual look inside packages in transit to reveal their
contents and value. The application establishes a central
administrator who can distribute inventory, packing and shipping
instructions to appropriate departments. Ultimately, not only the path
of the package but also the contents and value of the package will be
tracked.
the information business
"The information about a package is as important as
the delivery of the package itself."
• FedEx chairman and founder Fred Smith had that vision
in 1979, and it remains the heart and soul of the FedEx
technology story. It's not about bits and bytes, but about
delivering information, and it has revolutionized the way
business is conducted in a global economy.
• FedEx Corporation is a world leader in technology,
setting the industry standard for efficiency and customer
service. Its technological advances have always been in
response to customers' needs, anticipated future
requests and the demands of an information-driven
environment.
teaching and research
• Construction is in progress on the FedEx Technology
Institute at the University of Memphis, a state-of-the-art
facility designed to house an educational endeavor
teaching the newest technologies using the most
advanced learning techniques.
(http://fedex.memphis.edu)
• The facility will give faculty and students throughout the
university access to cutting-edge information technology
for learning and research. The primary objective is to
provide an environment that produces graduates
prepared for employment in the rapidly changing world of
the Internet and information technologies.
• FedEx Center of Cycle Time Research (under
construction) (http://www.people.memphis.edu/~cctr)
Recommendations for the future
• Continue to focus on customer service thru website
development.
• Better integration of data for internal and external
customers will continue to result in increased cost
savings.
• FedEx’s business proved to be very profitable, however
the airfreight industry has very low margins. The
inventory management and logistics businesses have
wider margins. Expanding in this market will enable
FedEx to better utilize its fixed assets.
• The outsourcing of corporate warehousing, ordering, and
shipping functions will provide a growing market for
FedEx to capture more revenues and improve profit
margins.
• The international market is growing at about twice the rate
of the domestic market, FedEx should continue to invest
in its international operations.
resources & reading materials
Chapters 8, 9 and 10 of textbooks
• http://www.fedex.com
• http://fedex.memphis.edu
• http://www.people.memphis.edu/~cctr