You are on page 1of 39

Ten Keys to Success in

Optimization Modeling
Richard E. Rosenthal
Operations Research Department
Naval Postgraduate School
Key #1: Communicate Early and Often

Mathematical formulation – kept up to date


Verbal description of formulation
Executive summary – in the right language
Mathematical Formulation
Index use

Given data (and units)


in lower case

Decision Variables (and units)


in UPPER CASE

Objectives and constraints


Verbal Description of Formulation

“Constraints ensure that one service facility is assigned responsibility for


each product line p.”

You wonder why I mention this, but look at our applied literature.
Non-mathematical Executive Summary

Jerry Brown’s Five Essential Steps:


• What is the problem?
• Why is the problem important?
• How will the problem be solved without you?
• How will you solve the problem?
• How will the problem be solved with your results, but
without you?
Key #2: Bound all Decisions

A trivial concept, too-often ignored

Remember all the formal “neighborhood” assumptions underlying your


optimization method?

Bob Bixby tells of real customer MIP with only 51 variables and 40
constraints that could not be solved… until bounds were added, and then it
solved in a flash.
Bound all Decisions
• Optimization is an excellent way to find data errors, but it really exploits
them

• Moderation is a virtue

• Bonus! You never have to deal with the embarrassment (or the theory) of
unbounded models.
Key #3: Expect any Constraint to
Become an Objective, and Vice Versa
Real-world models are notorious for multiple, conflicting
objectives
Expert guidance from senior leaders is often interpreted as
constraints
These “constraints” are often infeasible
Discovering what can be done changes your concept of what
should be done
Contrary to impression of textbooks, alternate optima are the
rule, not the exception.
Key #4: Sensitivity Analysis in the Real World Is
Nothing Like Textbook SA

LP Sensitivity Analysis, Textbook Style

Disappointing in practice because theory creates limits.


•  

Textbooks have sleek algorithms for one modification at a time, all


else held constant, e.g.,

minimize Sj cj Xj + dXk

Not very exciting in practice. So why is this stuff in all the


textbooks? What is worth talking about?
Sensitivity Analysis, Practitioner Style
Large-scale LP for optimizing airlift -- multiple time-space
muticommodity networks, linked together with non-
network constraints .
Initial results on realistic scenario: only 65% of required
cargo can be delivered on time.
Analysis of result revealed most of the undelivered cargo
was destined for City A from City B, so.. what if we
redirect some of this cargo to City A’?
Sensitivity analysis: add ~12,000 new rows and ~10,000
columns... on-time delivery improves to 85%.
Key #5: Bound the Dual Variables
Elastic constraints,
with a linear (or piece-wise linear) penalty per unit of violation,
bound the dual variables

“I’m willing to satisfy this restriction (constraint),


as long as it doesn’t get too expensive.
Otherwise, forget it;
I’ll deal with the consequences”
Key #6: Model Robustly
Your analysis should consider alternate future scenarios, and render
a single robust solution.

There may be many contingency plans,


but you only get one chance per year
to ask for the money to get ready.
Key #7: Eliminate Lots of Variables

Big models get to be big through Cartesian products of indices

Find rules for eliminating lots of index tuples before they are
generated in the model

Sources of rules: mathematical reasoning and common sense based on


understanding of the problem

You can often eliminate constraints too!


Example 1 of Variable Elimination

XD(a,i,r,t) = # of type a aircraft direct delivering cargo for customer i on


route r departing at time t

Allow variable to exist only if


Route r is a direct delivery route from customer i ’s origin to i ’s
destination
Aircraft type a is available at i ’s origin at t
Aircraft type a can fly route r ’s critical leg
Aircraft type a can carry some cargo type that customer i demands
Time t is not before i ’s available-to-load time
Time t is not after i ’s required delivery date + maxlate – travel time
Example 2 of Variable Elimination
Ann Bixby and Brian Downs of Aspen Technology developed
real-time Capable-to-Promise model for large meatpacking
company

One of their major efforts to bring solution times down low


enough was variable elimination.
Key #8: Incremental Implementation
In a complex model, add features incrementally. Test each new
feature on small instances and take no prisoners.

When new features don’t work, there is either a bug to be fixed or a


new insight to be gained. Either way, treasure the learning
experiences.
Incremental Implementation
Eliminate variables corresponding to airlifters switching from
long-haul to shuttle status, if there are no foreseeable shuttle
opportunities.

Euro

US SWA FOB

Feature tested with small example: removing the


option to make a seemingly foolish decision actually
caused degradation of objective function.
What happened?
Key #9: Persistence

“Any prescriptive model that suggests a plan and then, when used again,
ignores its own prior advice…
… is bound to advise something needlessly different, and lose the faith of
its beneficiaries.”

Jerry Brown
Illustration of Persistence
There are initially 8 customers to serve. We must choose serving site and
equipment type.
Results for Initial Case with 8 Customers
Customer Eqpt Type Site Distance
Cust01 Eqpt-01 HHH 353
Cust02 Eqpt-01 HHH 724
Cust03 Eqpt-01 HHH 773
Cust04 Eqpt-01 YYY 707
Cust05 Eqpt-01 YYY 719
Cust06 Eqpt-03 RRR 495
Cust07 Eqpt-01 HHH 442
Cust08 Eqpt-03 RRR 590
Illustration of Persistence
Just a moment after this solution is announced, two high-priority
customers call in. The model is rerun with the 10 customers.

There are not enough assets to cover all 10 customers.

The new solution requires a major reallocation of assets. Major


changes in the solution are highlighted.
Illustration of Persistence
Results for 10 Customers without Persistence
Customer Eqpt Type Site Dist (nm)
Cust01 Eqpt-01 HHH 353
Cust02 Eqpt-01 YYY 678
Cust03 Eqpt-01 YYY 703
Cust04 NOT SERVED
Cust05 Eqpt-03 RRR 705
Cust06 Eqpt-03 RRR 495
Cust07 Eqpt-01 HHH 442
Cust08 NOT SERVED
Cust09 Eqpt-01 HHH 353
Cust10 Eqpt-01 HHH 353
Illustration of Persistence

A persistent version of the model is run to obtain


a new optimal solution that discourages major
changes from the original announced solution.
Add to objective function: penalties on deviations
from original solution, weighted by severity of
disruption.
Illustration of Persistence
Results for 10 Customers with Persistence
Customer Eqpt Type Site Dist (nm)
Cust01 Eqpt-01 HHH 353
Cust02 Eqpt-01 HHH 724
Cust03 Eqpt-01 HHH 773
Cust04 NOT SERVED
Cust05 Eqpt-01 YYY 719
Cust06 Eqpt-02 RRR 495
Cust07 Eqpt-01 HHH 442
Cust08 NOT SERVED
Cust09 Eqpt-01 YYY 380
Cust10 Eqpt-01 RRR 363
Illustration of Persistence
Comparison of Solutions
Orig. Subsequent
With persistence? - No Yes
Customers 8 10 10
Objective function value
Customers not served 0 180.00 180.00
Original objective 206.67 151.75 155.20
Total 206.67 331.75 335.20
At a cost of 1% in the objective function, the persistent solution
causes no disruption to the announced plans, other than
substitution of the two new customers.
Key #10: Common Sense

• Heuristics are easy --- so easy we are tempted to use them in lieu of more
formal methods

• Heuristics may offer a first choice to assess a “common sense” solution

• But, heuristics should not be your only choice


Common Sense

A formal optimization model takes longer to develop, and


solve

But it provides a qualitative bound on each heuristic solution

Without this bound, our heuristic advice is of completely


unknown quality

This quality guarantee is key


Common Sense
It’s OK to use a heuristic,
but you should pair it with a traditional, “calibrating” mathematical model

With no quality assessment,


you are betting your reputation
that nobody else is luckier than you are
10 Keys to Success in Optimization Modeling

#1 Write formulation, #6 Model robustly


communicate with execs
#7 Eliminate variables – avoid
#2 Bound decisions generating them when you can

#3 Objectives and constraints #8 Incremental implementation


exchange roles (alt. optima
likely) #9 Model persistence

#4 Forget about sensitivity #10 Bound heuristics with


analysis as you learned it optimization

#5 Elasticize (bound duals)


MODELLING PROCESS
System analysis
Introduction Processes
Model
Space
Bounding Systems Time
Definition Niche
Elements
Word Models Impacts
Factorial
Alternatives Confounding
Separate
Hypotheses Combinations

Data Plotting
Outliers
Modelling Analysis Test

Choices Estimates
Validation

Conclusion
Integration
Communication
HYPOTHESES MODELLING PROCESSES
Decision Table
Relevance Processes Relationships

Variable Linkages Linear


Species Non-Linear
Impacts
Sub-systems Interactive

A hypothesis (from Greek ὑπόθεσις; plural hypotheses) is a proposed explanation for an


observable phenomenon. The term derives from the Greek, ὑποτιθέναι – hypotithenai
meaning "to put under" or "to suppose." For a hypothesis to be put forward as a scientific
hypothesis, the scientific method requires that one can test it.
Scientists generally base scientific hypotheses on previous observations that cannot be
satisfactorily explained with the available scientific theories. Even though the words
"hypothesis" and "theory" are often used synonymously in common and informal usage, a
scientific hypothesis is not the same as a scientific theory – although the difference is
sometimes more one of degree than of principle.
30
HYPOTHESES

Hypotheses of Relevance: Mengidentifikasi dan


mendefinisikan variabel dan subsistem yang relevan dengan
permasalahan yang diteliti

Hypotheses of Processes: Menghubungkan subsistem (atau


variabel) di dalam permasalahan yang diteliti dan
mendefinisikan dampak (pengaruh) terhadap sistem yang
diteliti

Hypotheses of relationships: Merumuskan hubungan-hubungan antar


variabel dengan menggunakan formula-formula matematik (fungsi
linear, non-linear, interaksi, dll)
Hypotheses of relationships
Determinants of the sophistication of SFA. In terms of specific
hypotheses:
1. Hypothesis 1: There is a positive relationship between
perceived strategic importance of sales decisions and level
of information orientation.
2. Hypothesis 2: There is a positive relationship between
organizational slack and level of information orientation.
3. Hypothesis 3: There is a negative relationship between
organizational control and level of information orientation.
4. Hypothesis 4: There is a positive relationship between
integration of IT and sales and level of information
orientation.
5. Hypothesis 5: There is a positive relationship between
perceived strategic importance of sales decisions and
integration of IT and sales.
6. Hypothesis 6: There is a positive relationship between
organizational slack and integration of IT and sales.
7. Hypothesis 7: There is a negative relationship between
organizational control and integration of IT and sales.
8. Hypothesis 8: There is a positive relationship between level
of information orientation and a count of the number of
types of results of sales campaigns that are measured.

Diunduh dari: http://www.sciencedirect.com/science/article/pii/S0019850107001149 ……………


Hypotheses of relationships
Path model of hypothesized causal
relationships among the lengths of roots
and mycorrhizal fungal hyphae, three soil
C pools and the percentage of water-stable
soil aggregates in a chronosequence of
prairie restorations. Fine roots are 0.2–
1 mm dia; very fine roots are <0.2 mm
dia.Single-headed arrows indicate direct
causal relationships and double-headed
arrows indicate unanalyzed correlations.
Numbers are path coefficients and
proportion of total variance explained (r2;
shown in bold italics) for each endogenous
(dependent) variable (n=49). The numbers
within ellipses represent the proportion of
unexplained variance [(1−r2)1/2] and, thus,
indicate the relative contribution of all
unmeasured or unknown factors to each
dependent variable

Diunduh dari: ……………http://www.sciencedirect.com/science/article/pii/S0038071797002071


VALIDATION MODELLING PROCESSES

Verification Critical Test Sensitivity


Analysis
Subjectives
Uncertainty
Analysis
Objectivities
Resources

Reason-ableness Experiments Interactions

Model verification and validation (V&V) are


essential parts of the model development
process if models to be accepted and used to
support decision making
Model validation is possibly the most important step in the model building sequence. It is also one of
the most overlooked.
Often the validation of a model seems to consist of nothing more than quoting the R2 statistic from the
fit (which measures the fraction of the total variability in the response that is accounted for by the
model).
ROLE OF THE COMPUTER
Roles
Speed
Introduction Data
Reasons
Algoritm
Manual
Comparison Calculator
Speed Computer
Techniques
Implication Errors
Plotting Repetition
Waste Checking
9/10
Data Modelling
FORTRAN
Program BASIC
High level ALGOL
Algoritms Language Machine code
DYNAMO. Etc.
Information Special

Development 35
Conclusions
Programming
ROLE OF THE COMPUTER

DATA
Machine readable

Cautions Availability Format

Sampling
Punched card
Exchange
Paper tape
Format
Reanalysis Magnetic

Tape
Data banks
Disc
36
DATA

Data adalah kumpulan angka, fakta, fenomena atau keadaan atau lainnya,
merupakan hasil pengamatan, pengukuran, atau pencacahan dan sebagainya

terhadap variabel suatu obyek, …..
yang berfungsi dapat membedakan obyek yang satu dengan lainnya pada
variabel yang sama

Data adalah catatan atas kumpulan fakta.

Data merupakan bentuk jamak dari datum, berasal dari bahasa Latin yang berarti
"sesuatu yang diberikan". Dalam penggunaan sehari-hari data berarti suatu pernyataan
yang diterima secara apa adanya. Pernyataan ini adalah hasil pengukuran atau
pengamatan suatu variabel yang bentuknya dapat berupa angka, kata-kata, atau citra.

Dalam keilmuan (ilmiah), fakta dikumpulkan untuk menjadi data. Data kemudian diolah
sehingga dapat diutarakan secara jelas dan tepat sehingga dapat dimengerti oleh orang
lain yang tidak langsung mengalaminya sendiri, hal ini dinamakan deskripsi. Pemilahan
banyak data sesuai dengan persamaan atau perbedaan yang dikandungnya dinamakan
klasifikasi.

37
JENIS DATA
INTERVAL
• Komponen Nama
NOMINAL • Komponen Peringkat (Order)
• Komponen Nama (Nomos)
• Komponen Jarak (Interval)
• Nilai Nol tidak Mutlak

ORDINAL RATIO
• Komponen Nama • Komponen Nama
• Komponen Peringkat (Order) • Komponen Peringkat (Order)
• Komponen Jarak (Interval)
• Komponen Ratio
• Nilai Nol Mutlak 38
REFERENSI

Russell L. Ackoff (1999) Ackoff's Best: His Classic Writings on Management. (Wiley) ISBN 0-471-31634-2
Russell L. Ackoff (2010) Systems Thinking for Curious Managers. (Triarchy Press). ISBN 978-0-9562631-5-5
Béla H. Bánáthy (1996) Designing Social Systems in a Changing World (Contemporary Systems Thinking). (Springer) ISBN 0-306-45251-0
Béla H. Bánáthy (2000) Guided Evolution of Society: A Systems View (Contemporary Systems Thinking). (Springer) ISBN 0-306-46382-2
Ludwig von Bertalanffy (1976 - revised) General System theory: Foundations, Development, Applications. (George Braziller)
ISBN 0-807-60453-4
Fritjof Capra (1997) The Web of Life (HarperCollins) ISBN 0-00-654751-6
Peter Checkland (1981) Systems Thinking, Systems Practice. (Wiley) ISBN 0-471-27911-0
Peter Checkland, Jim Scholes (1990) Soft Systems Methodology in Action. (Wiley) ISBN 0-471-92768-6
Peter Checkland, Jim Sue Holwell (1998) Information, Systems and Information Systems. (Wiley) ISBN 0-471-95820-4
Peter Checkland, John Poulter (2006) Learning for Action. (Wiley) ISBN 0-470-02554-9
C. West Churchman (1984 - revised) The Systems Approach. (Delacorte Press) ISBN 0-440-38407-9.
John Gall (2003) The Systems Bible: The Beginner's Guide to Systems Large and Small. (General Systemantics Pr/Liberty)
ISBN 0-961-82517-0
Jamshid Gharajedaghi (2005) Systems Thinking: Managing Chaos and Complexity - A Platform for Designing Business Architecture.
(Butterworth-Heinemann) ISBN 0-750-67973-5
Charles François (ed) (1997), International Encyclopedia of Systems and Cybernetics, München: K. G. Saur.
Charles L. Hutchins (1996) Systemic Thinking: Solving Complex Problems CO:PDS ISBN 1-888017-51-1
Bradford Keeney (2002 - revised) Aesthetics of Change. (Guilford Press) ISBN 1-572-30830-3
Donella Meadows (2008) Thinking in Systems - A primer (Earthscan) ISBN 978-1-84407-726-7
John Seddon (2008) Systems Thinking in the Public Sector. (Triarchy Press). ISBN 978-0-9550081-8-4
Peter M. Senge (1990) The Fifth Discipline - The Art & Practice of The Learning Organization. (Currency Doubleday) ISBN 0-385-26095-4
Lars Skyttner (2006) General Systems Theory: Problems, Perspective, Practice (World Scientific Publishing Company) ISBN 9-812-56467-5
Frederic Vester (2007) The Art of interconnected Thinking. Ideas and Tools for tackling with Complexity (MCB) ISBN 3-939-31405-6
Gerald M. Weinberg (2001 - revised) An Introduction to General Systems Thinking. (Dorset House) ISBN 0-932-63349-8
Brian Wilson (1990) Systems: Concepts, Methodologies and Applications, 2nd ed. (Wiley) ISBN 0-471-92716-3
Brian Wilson (2001) Soft Systems Methodology: Conceptual Model Building and its Contribution. (Wiley) ISBN 0-471-89489-3

39

You might also like