Monday 23 February 2009

Spot the yield/revenue management!

When I was a child, there was a very popular series of cheap books called "I spy". In each one there were about 30 to 50 items that one might see in a chosen situation. "I spy in London", "I spy in a hospital", "I spy on the railways" and so on. There were brief explanations of each item, and a point score alongside; common items scored 5 to 20 points, rarer ones might score 40 or 50. (One of the latter, I recall, was seeing a valley with river, road, railway and canal alongside each other.) There was also space to note where and when the item had been seen. The books were educational and fun; some facts were in them which I never encountered elsewhere in my education.

I am thinking that there should be a similar book for yield/revenue management. "I spy RM". Increasingly, RM enters our daily lives. The text books tell us that it was largely a product of the 1980's and airline management, but the roots of it go far further back, to hairdressers who charge different prices on different days of the week, and hotels which have weekend rates. So where can one spot RM today?
(1) Transport especially in the UK on flights and rail journeys, but there are cases of coach and bus travel;
(2) Hotels -- certainly if you Google for the topic, it is RM in hotels that comes out top, possibly because hotel management is diffused compared with the centralised management of airlines.
(3) Restaurants and other food and drink outlets
(4) Theatres and (slightly) cinemas
(5) Delivery of goods -- this encounter was what prompted me to blog today; we decided to order from a supermarket for home delivery and discovered that there were different prices for different two hour delivery slots. (We only decided to use the service because we had a £10 off voucher, which more than covered the deleivery charge at any time.)

Are there other, even rarer cases of RM?

In IAOR, the research papers are indexed under Yield Management, not RM. I have found that the expression Revenue Management is used by accountants to describe budgeting, so I prefer to use Yield.

Monday 16 February 2009

Can the solution be implemented?

A story from the press today tells how the Paris Metro (underground rail network) has been forced to change its tickets. For over a hundred years, the Metro has used small tickets, 2inches long by 1inch wide, or probably 50 by 25 mm. Latterly these have had a magnetic stripe which is used in the automatic barriers. But the authorities have noticed that more and more often, the tickets become demagnetised while being carried by members of the public, and tickets carried by women are more likely to suffer than thse carried by men. The problem is the increasing use of magnets as clasps on handbags, magnets which are strong enough to keep the bag closed, and therefore strong enough to demagnetise a ticket. So, sometime soon, the Metro is joining other mass transport systems and using cards with RFID and chips.

There are links between this story and O.R.. First, the general one -- that the solution to an O.R. problem may have been appropriate once, but should be monitored to make sure that the setting remains the same. Hence the title of this blog: can the solution be implemented? Or are there some good reasons why the behaviour of some of the people involved has changed? Second, I hope that the O.R. team who work for the Metro have done some analysis of the time it takes to check a new-style card in order to be sure that the barriers can cope with the passengers using the system at peak times. Turnstiles are servers in a complex queue, and changes in service times affect the characteristics of the queue.

The story is here

Tuesday 3 February 2009

Optimising the sub-system or the system?


There was heavy snow in Southern England yesterday, and there were problems with the transport infrastructure. In London, all the buses were cancelled, flights were cancelled, trains delayed, roads congested and blocked by accidents. Inevitably, some people asked the question "Why?". "Why can other countries cope with heavy snow, while England cannot?"

"Why?" is, of course, the correct question for an O.R. analyst to ask. And, I am pleased to note, there were people prepapred to give an O.R. answer. It comes down to the words of today's title. Should one optimise the system or the sub-system? For the commuter whose journey has been disrupted, the sub-system is non-optimal because he or she believes that the optimal solution is for there to be no problems with the commuter's journey. But the speaker explained that optimising the system means that it is not economic to have expensive equipment and trained staff needed for one or two days every ten years. It was a matter of costs and benefits. Naturally, the speaker didn't use these words, but they were implicit. And then the speaker added that with increasing numbers of people able to work from home, the difference between optimising the system and optimising the sub-system became even greater.

The picture is of my back garden this morning.

Monday 2 February 2009

Commuting Behaviour

A further observation from my trip to London is how there is a great difference between the social mix of commuters on public transport between Exeter and London. In Exeter, relatively few male executives use public transport. The rail lines are reasonably busy, but few males use the buses for commuting. But in London, it is quite normal and acceptable to use public transport. Of course, the infrastructure is better, but there is a difference in psychology as well. I wondered what research had appeared about the popularity of public transport commuting in cities of varying sizes.

It is relevant to O.R. because anyone working on transport models needs to remember the acceptability and feasibility of different modes of transport in different places. Behavioural psychology has its place in operational research.

Google turned up some figures for the United States. The comments are interesting. I wonder how much correlation there is between population and these percentages? And is there a similar set of data for other countries? And are there geographical effects to take into account?

RAW DATA: PUBLIC TRANSIT USE

Top 5 Cities for Public Transit Use

  • 54.63% New York, NY
  • 37.72% Washington, D.C.
  • 32.66% San Francisco, CA
  • 31.65% Boston, MA
  • 25.92% Philadelphia, PA

With the exception of Washington , D.C. , every city here grew up in the horse and buggy days, with streetcar rail systems. The District of Columbia is part of the 1960-70s "graduating class" of newly subway-enabled cities, along with Atlanta (MARTA) and San Francisco Bay Area cities San Francisco and Oakland (BART). BART reaches regional airports, commuter rail systems CalTrain, ACE and Amtrak, and someday it may even roll down to suburban San Jose . Atlanta is planning to extend MARTA with its back-to-the-future PeachTree Street Trolley and improved bus service.

Bottom 5 Cities for Public Transit Use

  • 1.07% Fort Worth, TX
  • 1.03% Tulsa, OK
  • 0.97% Oklahoma City, OK
  • 0.54% Virginia Beach, VA
  • 0.40% Arlington, TX

Surprise! These southern cities would benefit from re-installing light rail systems. Adding rail would provide residents relief from high gas prices -- and improve these cities' economic competitiveness. With air-conditioning thrown in, light rail would also provide relief from summer humidity.

When should one use O.R.?

Last week I had to go up to London (why is it always "up to London"?) for a meeting. Three and a half hours in a train to get there, three hours back, and two hours in the meeting. Every time I visit the capital, I am struck by the differences between Exeter (population 110,000) and London (population 7.5 million). This time my attention was caught by the advertisements on the London Underground (OK, I should get a life!). Several stations now have adverts projected across the tracks. But, the advert which caught my eye was on the walls of the escalators, where flat panel screens have replaced some of the static adverts. The advert was for T-mobile, and featured the company's recent TV advert, of a pre-arranged, choreographed flash-dance at Liverpool Station. "Life is for sharing" was the punch line.

Back home, I found the advert in full on Youtube, and also the director's background story behind an amusing and challenging project. Having enjoyed them both, I wondered whether there would be any place in the making of the advert for O.R.. How could "The Science of Better" make the production "Better"?

I suppose that the main traditional O.R. techniques that could have been used would be forecasting and scheduling. Forecasting to make sure that there were enough resources at the right time and the right place. Scheduling to make sure that resources were used as well as they could be. But there is actually little incentive to use O.R. in such a project. Either the people were ready for the advert on the morning of January 15th 2009, or they were not. There's no 95% confidence. It is a one-off event. And there is probably the answer. The director and his team worked to a deadline, made decisions as they were necessary and used a great deal of experience and common sense. And it worked.

Maybe if the director had to make such a film every two months for the next two or three years, then there would be a place for analysis and improvement, but that is not how the advertising world works. Perhaps an analyst can offer suggestions for improvements next time ....

And the other problem with the idea of scheduling is that you are dealing with people. It is often said that many O.R. techniques work better with machines and inanimate objects than when people interact!