Thursday, 27 December 2012

Where do buses go? Part 2

In January, I asked the question,  "Where do buses go?" and mentioned a new housing estate in Exeter which had no bus service.  I questioned what principles and guidelines the bus company would follow in deciding to provide a regular service to such a large new development, leading on to questions about how OR might contribute to the information about such decisions.  

  I didn't name the new estate; so I can't claim any responsibility for the fact that earlier this month, the bus company announced that it would introduce a bus service to the new estate - it is the "J" service on the city bus map.   Here's a link to the news story

 On the other hand, this month has seen controversy in Devon, because one of the services between Exeter and other towns in the county has been rerouted, and no longer serves the intervening villages regularly.  The general manager was on the radio, and said that the withdrawal from the villages was in response to requests to make the journey time between the towns and Exeter a little shorter.  That management team had not received comments from the villagers - who had (naturally) assumed  that their service would continue (People who are satisfied do not comment on their service!)

After that earlier blog was published, I had a reminder that there is a 40 year-old paper on OR's contribution to bus scheduling in Delhi, published in the New Scientist magazine, (Jon Tinker, "How Delhi made the buses run on time" (New Scientist p64-66, 9 January 1975) which has been reproduced in "Selected readings in operational research for developing countries" edited by G. M. Luck and G. Walsham (1972).  If you find this slim paperback book for sale on the internet, the price may surprise you.  (Today I found two copies, one at £75, the other at £125)  

In addition, a great deal of OR work has been done by bus companies using OR for crew and vehicle scheduling.  Beijing and London used a lot of OR for transport scheduling for their Olympic Games in 2008 and 2012.  But my question was about decisions about routes, not their schedules.

Saturday, 22 December 2012

Copper in Devon - follow up

I wrote about the unforeseen collapse of the copper industry in Devon in a recent blog.  Obviously I am not the only person who has been thinking about the way that external factors can affect a large industry, or even an economy.  I was listening to the BBC radio programme ("More or Less") about mathematics and statistics and heard Nassim Taleb explaining his concept of anti-fragility and the black swan.  (According to Wikipedia: He advocates what he calls a "black swan robust" society, meaning a society that can withstand difficult-to-predict events. He proposes "antifragility" in systems, that is, an ability to benefit and grow from random events, errors, and volatility.)

Although it might have been difficult for the copper mine owners to find ways of benefiting from the development of new sources of copper ore, perhaps there were ways that their industry could have been made anti-fragile.  And would OR have helped do so?

Sunday, 16 December 2012

Traffic control systems - multiple objectives

I posted a comment to Laura McLay's blog on traffic congestion the other day, which reminded me about the way that the traffic flow in Exeter is managed by an adaptive computer program called SCOOT.  SCOOT measures traffic flow and queues via a number of sensors, set under the tarmac around the city.  It is programmed to recognise surrogates for congestion (traffic speed, queues reaching certain thresholds, etc) and then uses a set of criteria to decide when the traffic signals should change.  Because these measures are surrogates, there are times when the information is incomplete and even misleading.

Traffic management presents an everyday example of multiple, conflicting, objectives, as well as illustrating how surrogate measures are sometimes the only ones available to a decision-maker.  And as OR people, we need to remember that the world is messy, and we cannot measure everything!

It is a decade or more since I used SCOOT as a class example, so I may be excused for forgetting it when I was faced with a "I wonder why ...?" thought.  I regularly pass through the junction below on my way home from church.  I travel from north to south from Polsloe Road into Barrack Road crossing the two sets of lights at Heavitree Road and Magdalen Road - which are about 50 metres apart at this point. (Ignore the label saying Newtown Ward - Magdalen Road continues in a straight line, as it has done for at least 2000 years (the Romans used it)).

A junction in the eastern suburbs of Exeter
I had wondered why, when there were half-a-dozen vehicles going north-south, and two or three going south-north, the lights changed before all the north-south traffic had left Polsloe Road.  The lights seemed to be controlled by the south-north queue.  This happened regularly.  Yet, when there were half-a-dozen vehicles going north-south, and a similar number going south-north, then the lights allowed both the north-south and the south-north traffic to clear the intersection.

Then I remembered SCOOT.  Heavitree Road and Fore Street, shown in orange, is a major route into Exeter.  Barrack Road feeds a great deal of traffic which turns left at Chadni (an Indian Restaurant) and continues flowing into the city.  (And there is obviously substantial flow in the other direction.) Therefore these two merging routes have a higher weighting inside the multi-objective algorithm than the flow from the lesser road that I was using to approach the junctions.  Polsloe Road suffers so that flow along the arteries of the city can flow more smoothly.

Multi-objective decision making is a wonderful process - but not for everyone!

How a picture helps - copper in Devon


Sometimes a diagram only needs a sentence or two to explain its message.  Take the diagram above.  The horizontal axis is divided into decades, the vertical axis shows production of copper (ore and metal) in each decade for Devon and Cornwall, the south-western counties of the UK.  What can you see?  There were three decades (1830s, 1840s and 1850s) when production was steady.  For the previous century, production had increased steadily.  Then it dropped dramatically, so that the production in the 1890s was less than in the 1720s.

What happened?  Did the many copper mines in Cornwall and Devon all run out of mineral at the same time?  That would seem unlikely.  If you were the owner of a copper mine in the 1860s, could you have foreseen the crash?  For a time, the Devon Great Consols Mine was the biggest copper mine in the world.  Its owners must have expected (forecast?) further profitable years.  What we have here is actually the consequence of world affairs, and not simply local ones.  The production of copper plummeted because vast new sources of copper were discovered in Chile, the USA, Australia and elsewhere, and it was cheaper to mine the element there.

I came across this diagram in "England's Landscape - The South West" (edited by Roger Kain).  It illustrates at least two lessons for OR people.  (1) The value of a simple diagram.  (2) Forecasting is a difficult process, because there may be external factors which have not been important before.

There are other examples of dramatic changes to time series as a result of some global event which affects the status quo.  Immigration from Eastern Europe following the collapse of communism there has had many effects on the population of Western Europe.  Nearer to Exeter, the number of students from east and west Africa coming to the UK to study mining and other subjects dropped suddenly following the end of apartheid in South Africa; it became much cheaper to study in Cape Town than in Cornwall.

Monday, 3 December 2012

Cyber Monday online

Today is forecast to be the big day for online Christmas shopping.  Families will probably have had their last pay cheque (check for those outside the English-speaking world) and will have spent the weekend browsing online before making purchases today.

So, how much will be spent?  According to the news story, about £320million today.  According to the headline, we will spend £222,222 per minute.  "About" has become "Exactly".  At least the headline didn't say that we will spend £222,222 and 22p

Wonderful precision!