spacer spacer
spacer

spacer

spacer !-->

Nov 16 11
Still hanging on…
by Matt Saltzman
spacer

Everyone’s doing wrap-ups and I have a few minutes before the last couple of sessions, so here goes (totally unstructured):

  • Things seemed to run phenomenally smoothly, compared to some of my memories.  Good job, INFORMS staff!
  • The BBQ was pretty passable.  Definitely among the better receptions in recent history.  (I know, BBQ is a matter of taste and regional bias so at least some folks will disagree vehemently.  De gustibus non disputandum est.)
  • It’s still not all that easy being green.  But Tuesday afternoon, Bjarni Kristjannson tweeted an announcement of a prototype mobile conference app. It’s kind of buggy and still has a ways to go before I would give up my printed bulletin (even as heavy as it is), but it’s a compelling proof of concept.
  • Big conferences have advantages and disadvantages.  Lots of interesting talks and activities at a large meeting, but people are scattered all across the area.
  • Surely we can do a better job of de-conflicting the schedule.  We’re OR professionals, after all.
  • I spend too much time in the exhibit hall and not enough time at sessions.  Or maybe not.  I got to talk to lots of interesting folks at the COIN-OR booth.  Hopefully, they will find our tools useful (and report back to us).

Safe travels home, y’all.

Comments Off
Nov 16 11
How did INFORMS 2011 Change your Perspective on Old Problems?
by John Angelis
spacer

One of the tricky aspects of INFORMS is knowledge management. After having absorbed the intellectual prowess of the brightest intellects in our field, one is left wondering about retention. Specific knowledge is buried rapidly under new knowledge. Quick quiz: can you remember three pertinent facts from the 8AM Monday session? It can be a bit disheartening.

However, what I appreciate about INFORMS is the cumulative effect of many different talks on one’s framework for approaching problems. I will give an example from my own experience and hope to hear yours as well. I am a member of the Technology Management Society (Shameless JOIN NOW plug) and attended many talks on innovation, entrepreneurship, new product development and technology evolution. I was pleased at how much my perspective was broadened.

Consider the classic problem of trying to find improvements to an existing solution, specialized to an innovation context. Some sort of mechanism (say, user-generated ideas through open innovation) provides you with a pile of potential improvements to your existing product or service. How can one sort through those improvements to find optimal (or satisficing) solutions? If you are trained in optimization, the answer seems simple: construct an objective function, note any constraints, and use an algorithm to find solutions.

What the conference helped me realize is how the innovation context creates a lot of interesting variations on traditional approaches, and how limited my prior perspective was.

Quick cautions: I am intentionally highlighting a problem that I am not working on so as to avoid self-promotion, but as such I am not an expert on all areas. This blog is being written to create discussion and ideas rather than serve as a complete (or accurate!) research document. Essentially, I am experimenting with using a blog format as a vehicle for addressing half-baked ideas, as I mentioned yesterday. Also, brevity.

First, in an innovation context, ideas must be judged by an authority (or authorities) to be cited as good, which takes time. As such, it may be more important to rapidly discard poor ideas, so authorities can judge only on ideas that have a higher probability of being improvements. The approach of discarding infeasible ideas vs. searching for feasible ideas is the same in many contexts…but it may depend on a larger debate on whether good ideas exist in a vacuum or are evolving in real time.

Second, the judgment of the authority itself can be perfect at the moment it is given, but due to fluctuations in markets and technology, firm capabilities, and other contextual factors, an option that once was an improvement is such no longer. Essentially, Type I and Type II errors may become correct decisions, and vice versa.

Third, the nature of the relationship between ideas that are actual improvements and the sources (users who provided the ideas) becomes important. Do improvements only come from a limited number of skilled users, or should the firm cast a wider net to include all sources? Essentially, it’s an explore-exploit trade-off, made more complicated by network effects of ideas and users.

Fourth, user learning  complicates the system. For example, given enough feedback from the authority, will users improve the quality of their ideas? But, that feedback takes time or accuracy away from judging. Thus, should users judge each other?

Some of these factors I once knew, or had seen in other contexts. Others were completely new to me. I apologize for any descriptive errors. I attempted to minimize overly technical explanations and citations. But I did this little exercise to start the conversation that I hope continues after we all leave Charlotte. I understand not all will want to share their new insights publicly in a comment, but I hope that you too can think of how many ways in how these talks have improved your perspective. Thanks for an excellent conference, and I look forward to further conversations via Twitter, email, and blogs.

1 Comment
Nov 16 11
What a Great INFORMS Conference and Community
by Anna Nagurney
spacer

As the conference draws to a close I wanted to thank the organizers for such a wonderful conference.

Although at first the size of it, in terms of the number of registrants (over 4,000) and the number of sites for talks and events from the Charlotte Convention Center to the various hotels may have seemed daunting, there was an intimacy to many of the sessions, business meetings, award ceremonies, and receptions.

In addition, the smiles and even hugs that greeted us brought a special warmth to this meeting.

The Charlotte conference brought us all together. I think that one of my former doctoral students at the Isenberg School of Management at UMass Amherst, who is now tenured at the University of Sydney and traveled all the way from Australia, said it best — It is about the people and the ones who mean so much to me. I had to come to the conference.

Wishing everyone safe travels back and until we meet again!

Comments Off
Nov 16 11
Wagner Prize and Final Thoughts
by Tallys Yunes
spacer

It’s almost time to go home, which means it’s time for my last post. I made the mistake of leaving my laptop in the suitcase I stored at the hotel, so I’m writing from my phone, and I apologize in advance for any typos and bad formatting.

I had a great time at the conference this year: I engaged in productive work with co-authors, attended high-quality presentations, and met so many new people! My network of contacts has increased dramatically. Above all, I was very impressed with the attendance at the regular sessions. Almost packed rooms were the norm. That’s very encouraging for both session chairs and speakers.

Yesterday, before visiting the NASCAR museum, I had a chance to learn about MDP models for ambulance dispatching, location, and relocation, and finally got to shake hands with Laura McLay. I enjoy seeing non-deterministic problems being reduced to equivalent, deterministic ones, and I’m looking forward to reading Laura’s paper when it comes out. (I also need to learn more about MDPs in general.)

After that, I met with some tweeps on my way to photograph some NASCAR cars (strange sounding expression), and had the pleasure of shaking hands with Bjarni (INFORMS VP of IT), also for the first time. Bjarni’s energetic attitude is going to take INFORMS to the next level, where it deserves to be, so I want to congratulate him on his success so far.

I attended the Wagner plenary presented by Intel and I was very impressed by the complexity of their problem and the way they devised a multi-technique approach to tackle it. If anyone had any doubts about the effectiveness of genetic algorithms in practice (I never had), Intel’s successful implementation should dispel the remaining doubts. Here are a couple of interesting quotes I picked up from the presentation:

“We can fit 6 million transistors in the period at the end of this sentence, but not this monster period on the projector’s screen; think of the period at the end of a sentence on a printed newspaper article.”

“Intel became a fifty-billion-dollar corporation using spreadsheets for a lot of its decision making!” (I feel better about my MBA teaching tools now :-)

Well, not any more. The integrated decision support system that they developed is gradually changing Intel’s decision making from distributed local views to a more global and collaborative approach, with increased what-if analysis capabilities. Congratulations to this year’s finalists, and especially to Intel!

I still have some interesting talks to see in sessions WD15, WD19, WE20, and WE24, but for those of you who need to head home early, have a safe trip! See you next year in Phoenix!

spacer

spacer

spacer

1 Comment
Nov 16 11
MD: Don’t make me wait!
by Barry List
spacer

As middle-age has overtaken me, I must confess that studies about healthcare interest me more and more. I go (mostly) to well-care visits; my kids deal with a variety of health annoyances; and of course, there’s the dog and the vet. So I find it reassuring that operations researchers are not only examining the big picture on the healthcare system and diagnostic issues in various diseases but are also paying attention to how long I’ll be in the waiting room.
This is not a problem that should be sloughed off as minor compared to true medical issues. At the Patient Flow session in the Healthcare Applications track this morning, a physician in the audience noted that most patients experience between 8 and 12 handoffs between medical staff. This takes its toll in patient dissatisfaction and irritability; it also wears down healthcare support staff, who turn over as they burn out. The system suffers.
That’s why a couple studies at the session caught my attention. In the first, Stephen Lawrence of the University of Colorado presented a paper he wrote with Linda Laganga and Michelle Samorani examining a few different ways that medical offices can plan for patients who don’t keep their appointments (an average of 20%) and others who come in at the last minute with a medical emergency.
One method for appointment scheduling with deterministic service and no-shows imagines eight slots. For the purposes of this model, it doesn’t matter how regular patients are scheduled. But when it comes to scheduling latecomers, you want to space them out – schedule the first overbook in slot 1, the second in slot 5, the third in slot 7, and the fourth in slot 2. Using this model, utility goes up as long as the office doesn’t exceed four overbooks.
In another, stochastic model that he presented, Prof. Lawrence envisioned a more fluid 12 slots, with wait times between patients shorter at the beginning of the day, longer in mid-day, and then again shorter as the day approaches its close. As you’re assigning slots, keep them away from one another until they fill to reduce the chance of overlap.
He previewed a session coming up this afternoon that looks at another approach, which takes into account patients’ history of punctuality. By gathering this data, you can schedule reliable patients early in the day and less reliable patients later. That’s a simple rule for a medical office to assimilate, and he notes that it improves performance measures substantially.
Yann Ferand of Clemson University presented another paper, which he wrote with a team from the University of Cincinnati and Cincinnati Children’s Hospital, which showed pity for emergency room patients at the lower end of the critical care triage – patients who often wait for long periods as those at greatest risk are treated first. With emergency room visits up 20% from 1995 to 2005, how can a hospital fast-track these lower priority patients?
Ferrand’s team did their best to gather data at a small suburban hospital about process steps and times (one questioner in the audience noted that recording data reliably as a patient is in extremis puts at risk the reliability of the data). They studied the number of process steps in triage, treatment, and discharge, built a simulation model, and evaluated potential changes, in particular introducing a fast track for low priority patients at a peak time of the day. Since limited physician hours represent the major bottleneck, they concentrated on taking the pressure off doctors.
They designed the supertrack with three extra beds and bumped it up to five (INFORMS veteran Doug Samuelson has suggested dispensing with beds for non-critical patients and freeing up doctors and nurses to move more easily from patient to patient). When possible, they added a ‘scribe’ to gather medical information from patients. And they reduced the number of steps for those needing less than critical care.
The result, especially on days when the number of patients is high, is that the length of stay in both the main track and the supertrack improves.
That’s something I’ll keep in mind the next time my wife or I slips on the ice while walking the dog on a winter night and gloomily imagines how long we’ll be waiting for treatment at our local ER.

Comments Off
Nov 16 11
INFORMS Lights Up Charlotte
by Polly Mitchell-Guthrie
spacer
spacer

Duke Energy building displaying INFORMS blue

spacer

Not from Broadway but College Street, at the Westin Hotel

Now, for something lighter, pun intended.

INFORMS lit up this town, and these two photos show at least physical proof. One is from the sign over the Westin Hotel, where many attendees stayed and many sessions were held. The other is from the Duke Energy building, which they lit up last night in INFORMS blue in honor of our descendance upon this city. It’s been fun. I’m now off to some of my last sessions.

1 Comment
Nov 16 11
Jim Orlin — final impressions from an impressionist
by Jim Orlin
spacer

It’s Wednesday, and I will soon be heading home.  Here are some final impressions of the conference, in no particular order.

  • Whereas 30 years ago, I wanted to spend my time talking to established figures in OR, this conference, I seemed to spend at least as much time talking to young researchers.   Talking to people of my own cohort often leads to talking of the past.  Talking to younger researchers  leads more often of talking of the future.
  • Question:  If networking is “to cultivate people who can be helpful to one professionally”, what is it called if you are the one who can help?  Is it “being networked”?
  • One of the advantages of being an “old timer” is that my opinions are automatically taken more seriously;  or perhaps, others are giving me the impression that they are taking my ideas seriously.  If it is the later, it still feels good.
  • INFORMS seems to be doing very well.  And I’m not saying this just because Rina Schneur (who did her dissertation under my supervision) has been an outstanding President of INFORMS.
  • “Analytics” is hot.  It’s a great opportunity for our profession.  It’s also an opportunity for other professions who compete with us.  So, we do need to keep sharp and constantly improve.
  • There are a large number of really smart young researchers.
  • The field of networks is amazingly popular.  I take no credit for its surge in popularity, but I am willing to ride the crest of the waves.
  • A colleague described two types of attendees of a conference:  impressionists vs. ‘detail lovers’.  (Actually, I forget the name of the second type, and inserted something similar.) The impressionists want to hear the high level aspects of a talk, and will fill in many details themselves.  If the talk leaves a sufficiently positive impression, they will seek out the paper.  The ‘detail loves’ expect more details in the talk, and are disappointed if the talks stay at a high level.  I am clearly an impressionist.  This is the first time that I ever felt a common attitude of mind with Manet, Monet, and Van Gogh.
  • The quality of the presentations at this conference were on average much higher than the quality 25 years ago.
  • The mix of t
gipoco.com is neither affiliated with the authors of this page nor responsible for its contents. This is a safe-cache copy of the original web site.