You are on page 1of 4

When do managers use quantitative methods? In England, in the years 1914-1915, F. W.

Lanchester attempted to
treat military operations quantitatively. He derived equations
1. The problem is complex relating the outcome of a battle to both the relative numerical
2. The problem involves many variables strength of the combatants and their relative firepower.
3. There are data which describe the decision environment
4. There are data which describe the value or utility of the  T.A Edison’s Statistics on Submarine Warfare
different possible alternatives
Thomas Alva Edison studied antisubmarine warfare, collecting
5. The goals of the decision-maker or her organization can
statistics and devising a war game. He analyzed zigzagging as a
be described in quantitative terms
merchant ship countermeasure and published findings on waiting
6. Workable models are available for these situations
times and call numbers in an electrical engineer's journal.
This helps managers make better decisions - Management
 A.K Erlang Early work in waiting time
Science/Operations Research

A. K. Erlang's 1917 work, "Solutions of Some Problems in the Theory


Early examples of Industrial Engineering Approaches
of Probabilities of Significance in Automatic Telephone Exchanges,"
1. Jethro – Moses’ father-in-law, credited with a treatise on introduced crucial waiting-time formulas for telephone traffic
organization principles in chapter 18 of the Book of theory.
Exodus  F.W HARRIS DEVELOPS THE FIRST INVENTORY
2. Charles Babbage – 1832, On The economy of machinery MODEL
and manufacturers
The economic lot size models in inventory control have a long
3. Frederick W. Taylor – (late nineteenth century) converted history, with the first published model attributed to Ford W.
industrial engineering into a profession; considered the Harris in 1915, despite G. D. Babcock's cubic equation.
father of scientific management  W.SHEWART APPLIES STATISTICAL INFERENCE
4. Henry L. Gantt – scheduling production Probability theory and statistical inference have been part of
management theory since 1924, with Walter Shewhart's
introduction of quality control charts. H. F. Dodge and H. G. Romig
contributed to the development of sampling inspection and
statistical sampling tables, while T. C. Fry contributed to the
statistical foundations of queuing theory through lectures and a
book.
 R.FISHER AND MODERN STATISTICAL METHODS
Sir Ronald Fisher's work on modern statistical methods is crucial in
understanding MS/OR development, forming the basis for most
applied statistical theory today.
 W. LEONTIFF DEVELOPS A LINEAR PROGRAMMING
MODEL
Wassily Leontieff, a Harvard professor, developed a linear
programming model for the U.S. economy, leading to numerous
military and industrial applications.
 H.C LEVINSON DOES MS/OR WORK IN THE DECADE
Horace C. Levinson, an astronomer, pioneered business MS/OR in
the US, applying scientific methods to business problems, studying
advertising-sales relationships, customer income relationships, and
purchasing patterns.

WORLD WAR II
 BLACKETT’S CIRCUS

Professor P. M. S. Blackett led Britain's early MS/OR group,


"Blackett's Circus," which included physiologists, mathematicians,
Early MS/OR -- PRE-WORLD WAR II astrophysicists, and military officers. Their mixed-team approach
demonstrated the value of barrage balloon deployment and
 Archimedes and the Siege of Syracuse "scattering" supply ships to minimize submarine attack damage and
deter aircraft attacks.
212 B.C., when the city of Syracuse employed Archimedes (who was  EARLY US OR GROUPS
then a man of 75 years) to devise means of breaking the naval siege
of the city, which was then under attack by the Romans. James B. Conant and Vannevar Bush were key figures in the
development of MS/OR in the US during WWII. They observed
 F.W Lanchester’s equation tested operations analysis groups in England in 1940 and 1942. General
Arnold recommended including operations analysis teams in Air
Force commands, with the first team assigned to England.
4 Sometimes MS/OR specialists forget to counsel managers on the
POST-WORLD WAR II limitations of the models they build, including the fact that many of
them have to be combined with judgment and intuition for effective
Operations research activity was valued by American military use.
leaders, leading to its continued use in various branches. The Army,
Navy, and Air Force continued operations research functions, 5 Often, managers forget to include an important constraint or
including the Operations Research Office and Project RAND. assign an incorrect value to a constraint.
However, it wasn't until bolder companies tried it with success and 6 Many MS/OR solutions are so complex that they are difficult to
word spread about its World War II accomplishments that civilian explain to managers in a way that builds support and confidence.
MS/OR gained significant momentum. 7 Many "real world" problems just don't have an MS/OR solution.

In 1947, George B. Dantzig developed the simplex solution to the How do you define a good MS/OR solution?
linear programming problem, which has become the standard
approach for various problems. 1 The solution should be technically appropriate. It should produce a
solution that works technically, meets the constraints, and operates
Management science/operations research today in the problem environment.
2 The solution should be reliable. It should work time after time
MS/OR activities have grown rapidly in business, government, and under the conditions for which it was designed.
private institutions. In the 1950s, the Operations Research Society 3 The solution should be economically viable. It should produce
of America (ORSA) and The Institute of Management Sciences value for the organization over what it costs to develop and it
(TIMS) were formed in the US. They published journals Operations should be seen as a wise investment of MS/OR talent.
Research and Management Science and collaborated on publications 4 The solution should be behaviorally appropriate. It should be
like OR/MS Today. In England, the Operations Research Club viable in its organizational setting, it should have the support of
formed the UK, which later launched the European Journal of management, and it must work well within the organization on a long-
Operational Research. MS/OR societies have been formed in many term basis.
countries, with the Decision Sciences Institute in 1971.
Quantitative methods to be discussed
Opportunities and shortcomings of the quantitative approach
Forecasting is an unavoidable responsibility of management. Faced
with uncertainty concerning the future, management looks to past
Opportunities behavior as an indicator of what is to come. Among the topics
treated are moving averages, exponential smoothing, trend-
1 MS/OR forces managers to be quite explicit about their adjusted exponential smoothing, trend line fitting, and the
objectives, their assumptions, and their way of seeing constraints. decomposition of seasonal data.
2 MS/OR quickly points out gaps in the data required to support
workable solutions to problems. Decision theory is concerned with making sensible decisions both
3 MS/OR permits us to examine a situation, change which decisions under conditions of complete uncertainty about future outcomes
are being made, and examine the changes without serious damage or and under conditions such that you can make some probability
excessive cost. statements about what you think will happen in the future. Methods
4 MS/OR forces managers to be very precise about how the are presented by which probability theory can be coupled with
variables in a problem interact with each other. financial data to generate valuable decision algorithms.
5 MS/OR makes managers consider very carefully just what
variables influence decisions. Inventory models make decisions that minimize total inventory
6 MS/OR lets us find a solution to a complex problem much more costs; these approaches can successfully reduce the total cost of
quickly than if we had to compute it by hand and often is the only purchasing inventory, carrying inventory, and being out of stock of
way we can solve large complex problems. inventory. Methods useful in dealing with discount evaluation, the
7 MS/OR lets us model a problem and its solution so that future joint ordering of items from the same vendor, and making inventory
solutions can be done by a computer, thus freeing management time decisions in the absence of complete information are also discussed
for decisions that require a more intuitive approach. at some length. Here we also develop a model that suggests that
being intentionally "out of stock" of an item may be a sensible
Shortcomings alternative.

1 Often, MS/OR approaches have to simplify the problem or make Linear programming is of value when a choice must be made from
simplifying assumptions to solve it, and thus produce solutions that alternatives too numerous to evaluate with conventional methods. It
have limitations. can be used to determine optimal combinations of the resources of
2 For problems that a manager must solve only one time, a firm to achieve a given objective or to allocate scarce resources
constructing a complex MS/OR model is often too expensive when optimally.
compared with other less sophisticated approaches.
3 Sometimes MS/OR specialists become so enamored with the Special-purpose algorithms are linear programming techniques
model they have built that they forget it doesn't represent the useful when working with certain specially structured problems. We
"real world" in which decisions must be made. illustrate the transportation method and the assignment method,
two approaches that are useful when management is confronted
with problems concerning the best distribution alternative or the  Most humans have an enormous memory, they forget a lot
optimal method of assigning operators to machines, accountants to and are not very precise in terms of what they do
audit teams, and even students to schools. These two methods are remember, but they can accommodate a wide variety of
included in a larger group of techniques known as network information from bits of data to complete complex
algorithms. thinking processes.
COMPUTERS
Integer programming, the branch and bound method, dynamic
programming, and goal programming are methods for choosing  Computers do only what their programmers tell them to do
among alternatives when answers may have to be found in whole in the form of instructions.
numbers, when the decision confronting management involves many  Computers can follow the rules but cannot learn
consecutive stages, or when organizational objectives have to be deductively except in the simplest logical situations.
stated in more than simple numerical terms.  Computers react consistently and, give or take electronic
problems, quite accurately.
Heuristics are sometimes described as "rules of thumb which work."  Computers can be programmed to follow complex,
An example of a commonly used heuristic is "Stand in the shortest interacting sets of rules; but even when the situation
line." Although using this rule may not work if everyone in the changes, they will continue to follow these rules toward
shortest line requires extra time, in general, it's not a bad rule to faulty outcomes.
follow. Many heuristics are very complex, and their formulation  Computers have flexibility only if someone has
involves a great deal of analysis. Heuristics tend to be effective in programmed it in; otherwise, they react identically
situations in which some of the more formal, mathematical MS/OR regardless of changed road signs.
techniques cannot be applied successfully.  Computers never forget; they have enormous memory in
terms of pieces of information, and they are very fast in
Simulation is a procedure that studies a problem by creating a retrieving this information.
model of the process involved in that problem and then, through a
series of organized trial-and-error solutions, attempts to determine ■HARDWARE is the "computer name" we give to all the electronic
the best solution to that problem. and mechanical HARDWARE devices that make up a computer.

Queuing theory studies random arrivals at a servicing or processing SOFTWARE is a term we use to refer to the instructions we give a
facility of limited capacity. Models allow management to calculate computer to perform. A set of instructions that "tells" a computer
the lengths of future limiting lines, the average time spent in line by how to perform a particular task is called a program. Any set of
a person awaiting future facility additions, and the service level programs that enables the computer to perform its functions is
waiting for costs and operating costs capacity that minimizes the called software.
sum of waiting for costs and operating costs
CLASSIFYING MANAGEMENT INFORMATION SYSTEMS IN
Network scheduling enables managers to cope with the complexities TERMS OF HOW OR WHERE THE PROCESSING IS DONE
involved in large projects; the use of this technique has
significantly reduced the time necessary to plan and produce There are at least five ways to classify management information
complex products. Techniques covered in the time needed include systems in terms of how or where the processing is done:
PERT (Program Evaluation Review Technique), CPM (Critical Path
Method), PERT/Cost, and scheduling with resource limitations. 1 Batch processing: In batch processing, the work is done
sequentially. Inputs (records, and so forth) are processed through
Markov analysis permits one to predict changes over time when the system in a predetermined or. der. Because large files are
information about the behavior of a system is known. Although the stored on tape or disk, the groups of transactions to be done must
best-known use of this technique is in predicting brand loyalties be gathered together and then sorted in the same order as the in.
(the brand behavior of consumers over time), Markov analysis also formation on the tape prior to processing.
has considerable use in other areas. 2 Real-time processing: Rather than starting with the first record
to be processed and finishing with the last, a real-time system uses
a nonsequential processing method. It provides access to any piece
HUMAN DECISION-MAKERS of information and finds that piece of information in the same
amount of time as any other piece; however, the response time must
 Humans have imagination, creative powers, judgment, and be kept small.
common sense. 3 Online processing: Online processing refers to equipment that
 Human decision-makers can learn from their experiences. operates under the control of the central processing unit described
 Human decision-makers are not always accurate; in Figure 1-5. You can locate online processing equipment in the same
sometimes their behavior is inconsistent. location as the central processing unit or in a remote location.
 Human decision-makers can see the "overall" problem (the Generally, online processing refers to processing using a terminal
forest, as it were); they can then see each subproblem as (input) that is remote from the central processing unit. Online
part of an overall scheme. processing can be used for batch processing or real-time
 Human decision-makers have flexibility (at least most of processing. When it is used for batch processing, it is generally
them do); when the road signs change, they can alter their referred to as remote batch processing or remote job entry.
behavior to optimize under the new situation.
4 Timesharing: Timesharing refers to an information system that 1. Are you a fairly sophisticated investor?
services many clients from one computer, these clients or users are YES
served simultaneously. 2. Is your current cash flow high?
5 Distributed data processing: Although this is not a true YES
distinction based on how the processing is done, it is a separate 3. Do you require a tax-sheltered investment?
category in terms of where it's done. In a distributed data NO
processing system, computing capacity is distributed throughout 4. Are you risk averse?
the organization (to functions, departments, and even individuals) NO
based on information needs.
Recommended decision: Buy high growth stocks.
Decision support systems (DSS) first joined the MS/OR vocabulary
Glossary
about 12 years ago. Different researchers and writers define DSS
differently. Some use a broad definition like "any combination of
Artificial intelligence An attempt to make machines think much like
MS/OR and computers which supports management decision
humans. Batch processing A processing mode in which the work is
making." It is useful, however, to be more specific in thinking of
accomplished sequentially.
DSS; thus, we can establish these characteristics of a DSS:

Blackett's Circus An early British MS/OR group under the


leadership of P. M. S. Blackett of the University of Manchester.
1 A system that provides management with quick access to
information that can be used in their decision-making.
DSI Decision Sciences Institute. Decision support system An
2 A system that has flexible databases.
information system that provides answers to problems and
3 A system that integrates MS/OR models with information
integrates the decision-maker into the system as a component.
processing software.
Expert system A system using MS/OR models that makes decisions
4 A system that helps management deal with unstructured problems
entirely without human intervention.
(problems that cannot be completely solved by an individual MS/OR
model, or by using an information systems report).
Heuristic A rule of thumb that works.
5 A system that improves the impact of management decisions by
extending the capability of managers who make those decisions.
Management information system A computer-based system that
6 A system that accommodates the intuition and judgmental
transforms data into information useful in the support of decision-
capability of management at all stages of decision-making.
making.
DSS, the decision maker makes the decision using the power of
Management science/operations research The systematic study of a
large databases and MS/OR models. In the section that follows, the
problem involves gathering data, building a mathematical model,
system itself makes decisions by itself.
experimenting with the model, predicting future operations, and
getting the support of management for the use of the model.
Artificial intelligence and expert systems

MS/OR Management science/operations research.


Whereas in DSS the system supports decision makers, utilizing
databases and MS/ OR models, the systems we will discuss in this
ORSA Operations Research Society of America.
section make decisions by themselves. Artificial intelligence, a
concept originally advanced by Nobel laureate Herbert A. Simon,
Sensitivity testing Altering the inputs to a model and watching what
can be defined as the attempt to make machines achieve thinking
happens to the output.
capacity much like that of humans. Dr. Simon stated that artificial
intelligence is an attempt to solve problems that:
Symbolic model An abstract model, generally using mathematical
1 Cannot be solved by the mathematics of optimization (standard
symbols.
MS/OR models)
2 Contain parts or components that cannot be quantified
TIMS The Institute of Management Sciences.
3 Involve large knowledge bases (things that are known)
4 Involve the discovery and use of alternatives as a part of the
solution process
5 Involve objectives and constraints that are often difficult to
specify in quantitative terms

■EXPERT SYSTEMS

Expert systems are a subfield of artificial intelligence which


concerns systems that reproduce the behavior of human experts.
Expert systems make decisions as an expert would in her field of
specialization, by mimicking her behavior as she solves the problem.
Many expert systems are written in advanced computer languages
which can handle symbols as well as numerical values.

You might also like