A sophisticated program develops from Bob Sehlinger's
Popular Unofficial Guide
Friday, July 18, 2003
by Mike Scopa, MousePlanet staff writer
The Unofficial Guide to Walt Disney 2003 by Bob Sehlinger offers
touring plans to its readers that are field tested with the objective
of giving guests the best chance of experiencing as many attractions as
possible while spending the least amount of time in line.
These plans are based on several factors, including theme park traffic
flow, attraction capacity, walking distance between attractions, and waiting
time data collected at specific intervals throughout the day and at various
times of the year.
In this session we will look into how the Unofficial Guide team
constantly seeks better ways to provide readers with the best guidelines
for touring the parks.
The Model
Unofficial Guide author Bob Sehlinger says the touring plans are
derived from “a combinatorial model that married the well-known assignment
problem of linear programming with queuing (waiting line) theory. The
model approximated the most time-efficient sequence in which to visit
the attractions of a specific park.”
The preliminary touring plan derived from this model is then field tested
in the park using a test group who follow the plan, and a control group
touring without the plan who use their own judgment.
The results? When theme park attendance exceeded 48,000, the group touring
without the plans spent an average of 3 1/2 hours more time waiting in
line, and experienced 37 percent fewer attractions than did those who
used the touring plans.
Sehlinger says, “As sophisticated as our model may sound, we recognized
almost a decade ago that it was cumbersome, slow, and did not approximate
the 'perfect' touring plan as closely as we desired. Moreover, advances
in computer technology and science, specifically in the field of genetic
algorithms, demonstrated clearly that it wouldn't be long before a model,
or program, was created that would leave ours in the dust.”
Although the touring plans were very good, Sehlinger's team was driven
to produce better results. And in the mid 1990s, they put forth a plan.
The Plan
Sehlinger's team joined up with Len Testa, a scientist and programmer
who was working in the field of evolutionary algorithms. It helped that
Testa was also a theme park junkie. They combined their years of experience
and data with Testa's vision and programming expertise to spawn a state-of-the-art
program for creating near-perfect touring plans.
Sehlinger is proud of the software. “The most recent versions of
the software are protected through a series of pending patent applications,”
he said.
This program was constantly matched against the best touring plans, which
had been published in the Unofficial Guide. In 2002, during field
trials conducted during the busy spring break period, the new program
beat the best touring plan generated by the traditional Unofficial
Guide model by 90 minutes at the Magic Kingdom.
Sehlinger points out, “Touring plans generated by the Unofficial
model saved an average of 3 1/2 hours standing in line compared to guests
touring without the plan. Well, the new program saved an additional 90
minutes over that. Needless to say, we're very pleased, but believe us,
it wasn't easy.”
The Challenge
The big challenge in devising an effective and dependable touring plan
is to determine which of the many combinations is the optimum choice.
Sehlinger gives us an example. “For example, if we wanted to visit
Space Mountain, Pirates of the Caribbean and Splash Mountain as soon as
the Magic Kingdom opens, there are six possible ways to do so.”
Plan #1
- Space Mountain
- Pirates of the Caribbean
- Splash Mountain
Plan #2
- Space Mountain
- Splash Mountain
- Pirates of the Caribbean
Plan #3
- Splash Mountain
- Space Mountain
- Pirates of the Caribbean
Plan #4
- Splash Mountain
- Pirates of the Caribbean
- Space Mountain
Plan #5
- Pirates of the Caribbean
- Splash Mountain
- Space Mountain
Plan #6
- Pirates of the Caribbean
- Space Mountain
- Splash Mountain
“Some of these choices are obviously better than others.” Sehlinger
explains, “Since the queue for Space Mountain increases very rapidly,
it's best to see it first thing in the morning. For similar reasons it
would be better to see Splash Mountain before Pirates. In this example,
touring plan number 2 would probably save us the most time standing in
line. Touring plan 5 would probably result in the most waiting in line.”
Of course as more attractions are added to the list, the number of possible
touring plans grows very rapidly.
51,090,942,171,709,440,000 ways to tour the Magic Kingdom
Are there any math majors out there? This was getting a bit deep for
me. Bob explained:
“Adding a fourth attraction would result in 24 possible touring
plans, since there are four possible variations for each of the six touring
plans listed above. In general, the number of possible touring plans for
n attractions is n x (n-1) x (n-2) x....x 1 (don't let the mathematical
notation throw you. If we plug real numbers in its quite simple.). For
five attractions, as an example, there are 5 x 4 x 3 x 2 x 1 possible
touring plans. If you don't have a calculator handy, that adds up to 120
potential touring plans. For six attractions, there are 6 x 5 x 4 x 3
x 2 x 1, or 720 possible touring plans. A list of 10 attractions has over
three million possible touring plans.
“The 21 attractions in the Magic Kingdom One Day Touring Plan for
Adults have a staggering 51,090,942,171,709,440,000 possible touring plans.
That's over 51 million billion combinations, or roughly six times more
than the estimated number of grains of sand in the whole world. Adding
in complexities such as Fastpass, parades, meals, and breaks makes the
number of combinations grow further.”
Phew. Glad he said that.
The Program
He goes on to say, “The Unofficial Guide's new program uses
a sophisticated, custom-made computer program to create most of its touring
plans. We know of at least one other software tool that creates touring
plans: RideMax for Disneyland. We tested an early version of the software,
but found it lacked certain important features we needed, and many of
the touring plans produced were significantly inferior to those derived
from the original Unofficial Guide model.”
Sehlinger's team also evaluated several commercially available software
packages wanted the software to work on a variety of computing platforms,
from multi-million dollar supercomputers, to ordinary personal computers,
as well as the World Wide Web and small personal digital assistants (PDAs)
similar to the Palm Pilot but no off-the-shelf software they examined
could do all these things.
The Unofficial Guide Touring Plan software program contains two
state-of-the art algorithms that allow it to quickly analyze tens of millions
of possible touring plans in a very short time. The software can analyze
Fastpass distribution patterns at all the attractions, for example, and
suggest the best times and attractions to use Fastpass. The software can
also schedule breaks throughout the day, say, to take a nap. If you're
going to eat lunch in the park, the software can suggest restaurants close
to where you'll be around lunchtime that will minimize the time you spend
waiting in line.
The Human Factor
The new program, however, is only part of what's needed to create a good
touring plan. Good data is also important. In a little more than a year,
Sehlinger's team of researchers made half a dozen weeklong trips to Walt
Disney World to collect data.
At the Magic Kingdom, for example, a group of researchers recorded the
estimated wait time at every attraction, show, Fastpass booth and restaurant
in the park, every 30 minutes, from park opening to park closing. On a
typical day the researchers walked about 18 miles and collected around
500 pieces of data each.
A typical route would have a researcher start at the Swiss Family Treehouse
in Adventureland. After collecting data on all of Adventureland, the researcher
would continue on to the attractions and restaurants in Frontierland.
After that came Liberty Square, then finally through half of Fantasyland,
before walking back to the Swiss Family Treehouse for an eight-minute
break before starting the next round of data collection. A platoon of
other volunteers collected data in the other half of the park.
How much did these researchers commit to their research? Just listen
to Sehlinger.
“One volunteer, conscientious to a fault, took his eight minute
break in the same spot every day, directly under the Tomorrowland Transit
Authority. He endured the same prerecorded TTA announcement (“Paging
Mr. Morrow, Mr. Tom Morrow”) every 23 seconds, for six days. We hear
his therapy is going well.
“All of the researchers endured blisters, heat and run-ins with
strollers. Their persistence and determination to collect the data was
truly astounding. Christine and Mike Testa, Linda and Rob Sutton, Guy
Garguilo, Mais Homsi, and Kenny Cottrell supervised on-site data collection
and did a bang-up job.”
So how good are the new touring plans in the Guide? The computer
program the Unofficial Guide uses typically gets within about 2
percent of the optimal touring plan. To put this in perspective, if the
“perfect” Adult One-Day touring plan took about 10 hours on
average to complete, the touring plan in the Guide would take about
10 hours and 12 minutes. Since it would take about 30 years for a really
powerful computer to find that “perfect” touring plan, the extra
12 minutes is a reasonable tradeoff.
Next Time
Now that we know all about how touring plans may be put together, we
should probably talk about how to make the best use of the Walt Disney
World transportation system. That's our focus for next time.
Class dismissed
|
There were many contributors to Unofficial
Guide Touring Plan software program. Among those involved were Dr.
Albert C. Esterline of North Carolina A&T State University, Dr. Gerry
V. Dozier of Auburn University, and Dr. Nikoloas Sahinidis and his graduate
students at the University of Illinois at Urbana-Champaign who have published
a number of exceptionally helpful studies. Other notable contributors
include Chryssi Malandraki, Ph.D., from United Parcel Service and Robert
Dial, Ph.D., from the Volpe National Transportation System Center.
Mike Scopa first visited Walt Disney World almost 30 years ago. Planning a trip was simple back in the 1970s, with only the Magic Kingdom and a few Disney-owned resorts in Orlando. Over the past 11 years, Mike has been perfecting his WDW trip-planning skills as he has hosted chats and bulletin boards about Disney for a Fortune 100 company. Mike brings his experience to MousePlanet in a series of lessons to help you with all the phases of planning a WDW trip. Mike pays special attention to all the details that ensure your family has the best possible time at the Happiest Place on Earth. You can contact Mike here. Here are trip reports that Mike has written that are part of MousePlanet's archives: Michael Scopa -- August 1999 -- Walt Disney World (CSR) Michael J. Scopa -- July 1997 -- Walt Disney World (WL/CBR) Mike Scopa -- July 1994 -- Walt Disney World (WL / CBR) Also, don't miss Lani Teshima's column, “The Trip Planner” for more travel planning information. Get the latest info about the resort at “Park Update: Walt Disney World.” 
|
