I’d be willing to bet that if you scanned your environment right now and pondered the state of your company, you’d be able to provide several splendid examples of less-than-stellar decisions some of your contemporaries have managed to inflict on the entire organization. Sometimes the reasons for having done so are quite obvious, sometimes not. What is certain, though, is that as the number of input variables affecting the final decision increases much beyond two or three, the human brain loses its ability to justify calling ourselves homo sapiens. In the real world, the amount of data to be evaluated can be overwhelming. Perhaps the data are valid, but even so, it’s probably incomplete, messy and appears to differ little from a series of random numbers or facts. There must be a way to make sense of that hash in the industrial world.
It can be done. You’ve heard of the electronic trading systems used in the financial markets. Skilled traders systematically reduce their experience and intuition to a set of rules that a computer executes to return clear, unambiguous buy and sell signals much faster than a human can. Sometimes the computer itself initiates the trades. It’s the closest thing we’ve got to a money machine.
Join me for a leap into that digital morass we call the Web in search of practical, zero-cost, noncommercial, registration-free resources that ought to help you when you’re “the decider.” Remember, we search the Web so you don't have to.
An easy read
So, what exactly do we mean by the term decision support system (DSS)? You can bet it involves a computer. Beyond that, it would be better to pay a visit to academia and talk to the experts. In this case, I’d suggest you contact Marek J. Druzdzel and Roger R. Flynn from the Decision Systems Laboratory at the School of Information Sciences and Intelligent Systems Program at the University of Pittsburgh. I hesitate to refer to them as prototypical ivory-tower types because they’ve written a 15-page paper titled “Decision support systems” that avoids any use of the abstract, esoteric and abstruse, which should make it perfectly comprehensible to anyone reading this column. The paper explains the characteristics of decision problems and how computer programs can support decision making. It goes on to detail the various DSS components and the role they play in decision support. The authors introduce an emergent class of something called normative systems and decision-analytic DSS. Finally, they review issues related to DSS user interfaces and the importance they have to the quality of any computer-assisted decision. You might as well decide to wander over to www.pitt.edu/~druzdzel/psfiles/dss.pdf and spend quality time with the document. It will be worth it.
You no doubt have heard the old joke about the guy who goes to the optometrist for a new set of spectacles and, naturally, asks about price. The optometrist tells him it will be $75. When the stunned guy hesitates, the optometrist quickly adds, “per lens.” When the guy, now in shock, still doesn’t respond, the optometrist says, “plus frames.” Capitalism is all about negotiating for the best price you can get, and the researchers in the medical profession know how to exploit a DSS to maximize the returns for practitioners. If you think this is a joke, the next Web article citation is titled “Willingness-to-pay utility assessment: feasibility of use in normative patient decision support systems,” by Flowers, Garber, Bergen and Lenert, all from the Stanford University School of Medicine. The link (www.ncbi.nlm.nih.gov/pubmed/9357621) takes you to an abstract of the article. That’s the best I could do. I found it impossible to finesse the digital morass to access the full text at any of the sites that mention the piece. The bottom line: You probably should be circumspect when talking to anyone in the medical field.
Come the fall, football season will be in full bloom, along with adjunct activities such as office pools and fantasy football leagues. Needless to say, if you’re going to put real cash into these ventures, you might want to consider the idea of using a DSS to slew the odds a bit more in your favor. One such single-purpose DSS is available, amazingly enough, on the Internet. Brandie Searle and Greg Alan developed the software and Automation Creations Inc., out of Blacksburg, Va., distributes it online. Although the software is free, you’ll need to subscribe if you want periodic data updates. I’d imagine that those who are sufficiently rabid about football will probably be able to wing it without the updates. That’s a personal choice you can make when you do an end run to www.pcdrafter.com to start betting smarter.
The financial markets in this country are said to be efficient. That means arbitrage, or buying a negotiable instrument at a low price in one market and selling at a high price in another, is nearly impossible, instant communication being what it is. Everything relevant that’s known about the stock or bond or whatever is already reflected in the price. Besides, there are thousands of people working on calculations and theories designed to beat the market. They’re all nibbling at the edges, never getting a big bite, and market prices reflect the truth about value. The collective action of so many players, a prediction market, constitutes a decision support system that indicates how much one should pay for a particular investment. But that’s not the only use for a prediction market. They’ve been applied to the Fed’s monetary policy, economic indicators, stock prices and more. They’re a sort of futures market that uses real money and has real payoffs; anyone can play, but it’s a winner-take-all game. If you’re interested in the power of prediction markets as a viable DSS, you should read “Prediction Markets as Decision Support Systems,” by Joyce E. Berg and Thomas A. Rietz at the Henry B. Tippie College of Business at the University of Iowa. This 15-page paper, available at www.biz.uiowa.edu/iem/archive/decisionsupport.pdf, provides background for the prediction market they established for the 2008 presidential election. You can learn more about it at www.biz.uiowa.edu/iem. Political junkies can now put their money where their mouths are.
Sometimes the data that fill a DSS can only be represented by a series of complicated, interdependent tables. If you have only one or two relevant variables, you could use a flat spreadsheet. With three variables, your spreadsheet morphs into a cube, which a spreadsheet could still conceivably handle. If you have four or more variables, you’re into hyperspace and analysis in that realm is going to require another approach. This is a good time to introduce the idea of OLAP, an acronym for online analytical processing, the software that proves useful for any multidimensional set of DSS data. There are several sources for OLAP software. In a completely capricious, arbitrary, random selection, I present for your consideration the www.filedudes.com. Pay them a visit and enter the term OLAP in the search box. Then, scroll down for an entry called OLAP + CHART MODELKIT 3.6 by Perpetuum Software. At that point, you can access a free download as long as you’re satisfied with a trial version.
A good collection
Decision Support Systems Resources is an evolving Web site that’s intended to be a repository of cutting-edge information about the topic. Launched during the summer of 1995 by Daniel J. Power, a professor of Information Systems and Management at the College of Business Administration at the University of Northern Iowa, Cedar Falls, the site has some good features. The important links are on the left side of the screen. For example, the folks engaged in designing and using decision support systems have, as you would expect, developed a certain jargon to better communicate their ideas. The “DSS glossary” link can have you speaking fluent DSS quickly. If you need to do a literature search to enhance your argumentative credibility vis-à-vis management, the “Articles” link gives you access to many free documents gathered from a variety of sources. One “channel” is directed at managers. A mere mouse click gives you a DSS FAQ and a selection of case studies. The other options require registration and cash. So, go to http://dssresources.com, but tread lightly.
Generic multi-attribute analysis (GMAA) is a type of DSS that tolerates incomplete input data. Lacking sufficient space to explain it in any more detail, I’ll refer you to www.dia.fi.upm.es/~ajimenez/GMAA, where you can learn more than you really want to know about the procedure. But, if what you read suggests an application in your plant, professors Sixto Ríos Insua, Alfonso Mateos Caballero and Antonio Jiménez Martín from El Departamento de Inteligencia Artificial de la Universidad Politécnica de Madrid have made it possible for you to download, for academic purposes only, a free copy of GMAA, the software that makes the analysis possible. This is no lightweight application; the user’s manual has 83 pages and you’ll have to spend some time digesting the content. One never knows; it might solve the maintenance crisis.
Closer to home
The purpose of this column is to convince you that decision support systems can be applied to activities in the industrial maintenance arena. A DSS is almost mandatory if one is to make sense of the gigabytes of CMMS data that describe the current and historical condition of the countless operating assets used in a multiplant operation. Making rational decisions when you’re drowning in data isn’t the easiest part of the day. Daniel J. Fonseca in the Department of Industrial Engineering, University of Alabama in Tuscaloosa, feels your pain. To demonstrate the practicality of DSS-enhanced maintenance, Fonseca developed a fuzzy expert system that supports reliability-centered maintenance (RCM) programs during the design of industrial chemical processes. You should read about his project in “Reliability Management Through Knowledge-Based Systems,” which is posted at http://ie.eng.ua.edu/research/MRC/Reliability_Thr_KnBsys.doc. Among other things, he cites four maintenance applications in which decision support systems have already been successful: overland communication cables, nuclear power plants, hydraulic power systems and optimizing maintenance routines. So, you see, it can and has been done. You can do it, too.
While you’re there, consider “A decision support system for machine replacement policies” that Fonseca wrote with two colleagues, Shital Shah and G. P. Moynihan. This one is at http://ie.eng.ua.edu/research/MRC/DSS_REPLACE_POLIC.doc.
E-mail Executive Editor Russ Kratowicz, P.E., CMRP, at [email protected].