Algorithms Key to Cheap Air Fare (Wired)

[Dave Farber <dave () farber net>]

------ Forwarded Message
From: GLIGOR1 () aol com
Date: Thu, 10 Apr 2003 12:15:50 -0400 (EDT)
To: dave () farber net
Subject: Fwd: Algorithms Key to Cheap Air Fare (Wired)

 


------ End of Forwarded Message

-------------------------------------
You are subscribed as interesting-people () lists elistx com
To manage your subscription, go to
  http://v2.listbox.com/member/?listname=ip

Archives at: http://www.interesting-people.org/archives/interesting-people/
> --- Begin Message ---
>
>
> From: Wdimitr () aol com
>
> Date: Thu, 10 Apr 2003 12:09:38 EDT
>
> Algorithms Key to Cheap Air Fare  
>
> By Louise Knapp 
>
> Story location: http://www.wired.com/news/technology/0,1282,58353,00.html
>
> 02:00 AM Apr. 09, 2003 PT
>
> Buying an airline ticket can be a nail-biting dilemma: buy now or hold off in 
> the hope that prices will
> drop? 
>
> A new computer program, aptly named Hamlet, seeks to answer the question "to 
> buy, or not to
> buy?" by predicting when air fares will be at their cheapest. 
>
> Airlines are renowned for raising and lowering fares in what seems to be a 
> random fashion. But,
> according to Hamlet's creator Oren Etzioni, computer science professor at the 
> University of
> Washington, there is a method in the seeming madness. 
>
> Etzioni said price variations are governed by algorithms developed by the 
> airlines and kept as closely
> guarded trade secrets. "The airlines have a bunch of high-powered consultants 
> who work with them
> to figure out the algorithm in the back room." 
>
> Hamlet works by studying past trends in price variation and establishing the 
> patterns. These are then
> used to predict future price variations. 
>
> "We have data-mining algorithms that run over past ticket prices available on 
> the Web, and we find
> patterns and make predictions from these patterns," Etzioni said. "These 
> suggest when the prices will
> fall and when they will rise." 
>
> To test the program, Etzioni fed Hamlet data from two nonstop flights over a 
> 41-day period. One of
> these flights was from Los Angeles to Boston, the other Seattle to 
> Washington, D.C. The data
> included six airlines on the routes. 
>
> "We started out by collecting data on prices from a large number of sites 
> every three hours, a total of
> eight times throughout the day," Etzioni said. "The prices don't change every 
> minute, so searching for
> this data several times a day is sufficient." 
>
> Hamlet proved very good at predicting the price variations, saving 607 
> simulated passengers a total
> of $283,904 on airline fares by advising when to buy and when to postpone 
> purchases. 
>
> Persuading the public of the program's worth may, however, be this Hamlet's 
> fatal flaw. 
>
> "What if they're wrong?" said Carol Jouzaitis, vice president for corporate 
> communications with
> Orbitz. "They can't see into the future. I am not sure that Orbitz would be 
> happy to support that --
> it's a risky proposition." 
>
> Sheryl Temple, public affairs manager for Horizon Air, agreed. "There are so 
> many variables that
> cannot be measured that determine air fares. I cannot even tell you what the 
> fares are going to be
> next week." 
>
> Travel agents appear equally skeptical. 
>
> "Trying to predict the price of airline tickets is like trying to predict the 
> weather in San Francisco --
> you never know what it will be. It depends on so many factors," said Nancy 
> Cathcart, travel agent
> with Buenaventura Travel, based in San Francisco. 
>
> Brad Hudson, manager of Travel Zone, another San Francisco agency, agreed. "I 
> would be really
> skeptical as to how a software program could tell when a ticket was going to 
> be cheaper, so I would
> have to be convinced, and then I would want to develop this for other things 
> too -- to predict the
> outcome of horse racing, for example," he said. 
>
> Etzioni is convinced, however, that Hamlet will prove its worth. 
>
> "Price changes are not random at all," he said. "They only appear highly 
> unpredictable to people who
> haven't studied the pattern, but airlines operate on these algorithms. They 
> are complicated, but the
> behavior is highly predictable." 
>
> William Maloney, executive vice president and operating officer of the 
> American Society of Travel
> Agents, agreed, but brought up another possible problem. 
>
> "If the airlines knew they were being tracked, they could counter-program 
> these to fool the system,"
> he said. 
>
> Etzioni, however, said if the airlines started playing with the prices to 
> trick Hamlet, they might also
> deviate from patterns devised to sell as many tickets as possible and, in 
> turn, lower their chances of
> selling a ticket. 
>
> Etzioni's next step is to look at different kinds of flights, including 
> business fares, international flights
> and multileg flights. 
>
> "Multileg flights will be complicated, as their price depends on more than 
> one leg of the journey --
> but we will be able, with more research, to predict these, too," he said. 
>
> Etzioni said Hamlet will be ready to make its debut in a year. 
>
> .
>
> © Copyright 2003, Lycos, Inc. All Rights Reserved.
>
>
> --- End Message ---