Saturday, 21 December 2013

How to leave Devon, if you need to!

Major sporting events have used O.R.to deal with the problem of getting spectators to and from the venue by public transport.  The organisers of the last three Olympic Games have all done so (and maybe others as well; not everyone involved with such planning publishes or publicises their work).  There is a problem that most readers will recognise - they have attended an event which finishes at a particular time, and the whole audience wants to get away at the same time.  A similar problem arises in airports where a jumbo jet lands and all its passengers spill out at the same time.  The capacity of the system for people needs to recognise that large peaks will occur.

One day several years ago, soon after I had written about the use of capacitated networks for transport planning, I had a phone call at work.  The caller asked whether what I had written might be applied to help plan evacuation routes.  It wasn't something that I had direct experience of, though I had been taught a little about the theoretical concept.  So there followed a conversation about the way that evacuation of an area in an emergency might use the capacity of the arcs in a network of roads, and I remember drawing attention to the problems of how the flow of people at road junctions would mean that there were capacities of the nodes as well, and some of the problems of measuring the data.  Towards the end of the call, I asked why the caller was interested in the subject.  That was when he revealed that he was drawing up plans in case of a nuclear accident affecting a town near one of the UK's nuclear power stations.  I never heard from that caller again, so I do not know what happened next.

The south-west of England, Devon (where I live) and the neighbouring county of Cornwall, are popular destinations for holiday-makers.  These days, the majority travel by car.  Unfortunately, the geography of the peninsula means that there are few major roads into the two counties. 
Devon; major roads in green, and the thick blue line is motorway.
The most popular change-over day is a Saturday, so these major roads become congested with the traffic of holiday-makers coming and going.  It's not really an evacuation, though sometimes it may seem like one.  And there is traffic in and out of the counties.  As the map shows, there are two major roads and a motorway connecting Devon with the country to the east.  I only travel when I really need to on a Saturday in the summer, as I know that there will be hold-ups!  The border between Devon and Cornwall also has three major roads; and for much of the border, there is a river (the River Tamar) to be crossed, with very few bridges. 

Soon after coming to live in Exeter, I heard a story that made an interesting claim about those holiday-makers.  If everyone who had to leave Devon and Cornwall on a given date were to set out at exactly the same time, then there would be no problems on the roads.  All the holiday-makers would be spread out uniformly across the roads.  There must be a grain of truth in the story, because in 1999, when the counties were crowded with people wanting to see the solar eclipse, the traffic flowed extremely smoothly.  Once the eclipse was over, then many people returned home, leaving their various vantage points within a short space of time after the event. 

In the January 2014 issue of the Journal of the Operational Research Society, DR Bish, HD Sherali and AG Hobeika have written a paper "Optimal Evacuation Planning using Staging and Routing" (JORS volume 65, issue 1, pp124-140, DOI 10.1057/jors.2013.3).  Their models include the concept of staging, where the evacuation plan for people in an emergency specifies both the route that they should take, and the time they should start to move.  There would probably be some resistance to such an idea in some kinds of an emergency because the natural reaction is flight.  Could it be used for summer Saturdays in Devon?  Unfortunately, the number of holiday-makers in Devon and Cornwall who want to change-over on a summer Saturday is too big for their model! 

Thursday, 19 December 2013

Spreadsheets - the good, the bad and the ugly

I recently mentioned in my blog that the last issue of journal OR Insight had just been published.  In the final issue there was an article "When spreadsheets go bad" by Andrew Tait and Kurt Richardson.  The article was easy to read, and thought-provoking.  It included some stories and one-line reminders about the tool that we all use every day.

  • When an operations analyst encounters a problem, he (or she) opens a spreadsheet.  Now he has two problems.
  • If you have a project that has a spreadsheet as a deliverable, go and suspend it now.  Yes, now.  Go on.
  • Spreadsheets should not be used for 'line of business' applications. In fact we'd go further - spreadsheets should only be used by the person who created them.
  • If you use a spreadsheet to store a database in a corporate setting, and it is doing more than manage your lottery syndicate, then an urgent review is in order.
  • If you use a spreadsheet to manage your accounts, and unless your spreadsheet is being used by your lemonade stall to see if you can afford that hamster, buy Quickbooks.
  • EUSPRIG maintains a list of horror stories about real-world losses through spreadsheet errors.  Don't read with the lights off.
  • How many business problems are actually a natural fit for a two or three dimensional grid of cells?
  • We'd even go so far as to paraphrase the National Rifle Association and say that spreadsheets don't create bad models - people create bad models.
  • Spreadsheets are actually excellent for prototyping, ... for exploring an idea, or for checking a theory, a model that you'll discard once you have an answer.  Prototypes are never used in production.
  • One wag suggested that spreadsheets would be used more effectively if the save function were removed.  We have some sympathy.

I realise that I have not blogged about spreadsheets since I changed the name of the blog.  I did blog under IAOReditor about another thoughtful article about spreadsheets.  You can find it here.
I recommend the OR Insight article to anyone who is teaching students of O.R. about model-building.

Saturday, 14 December 2013

Farewell to the journal OR Insight

This week the last issue of the journal OR Insight arrived.  It carried an editorial which explained:
OR Insight has been the practitioner-oriented journal of the Operational Research Society for many years. It was published in-house up to 2008, but since then it has been professionally produced .... OR Insight’s success continued after this change, leading to it being included in the Journal Ranking List produced by the Association of Business Schools and others.
The primary objective of OR Insight has been to publish articles that are readable, topical, relevant and of interest to managers, consultants and OR practitioners. It aimed to be an interesting and stimulating publication that would appeal not only to OR practitioners and consultants, but also to managers and others wishing to learn more about OR in practice. It sought to inform about not just the scope and potential of OR, but also developments in related areas.
For many years, I sat on the OR Society's publications committee (mostly as its minutes secretary).  I sat through many discussions about the need to make the work of OR practitioners accessible to a wider audience.  OR Insight was one of the outcomes.  Writers were encouraged to pitch their material to a non-technical readership.  The OR Society admired the style found in the American bimonthly ORMS Today, and the aims of the journal Interfaces.  (The latter was especially admired during Gene Woolsey's tenure as editor.)  Sadly, it was always hard to find writers who could pitch their writing appropriately, partly because it is easy to forget how little the outsider knows about OR, and partly because there was little kudos for writers (academic or practitioner) who published in such a journal.  When I retired, I didn't need the academic kudos, and published two articles in OR Insight, trying to get the style right.
  
OR is always going to be a hard subject to explain to the outsider.  Our work is so varied, our expertise is extensive, and our problems are so diverse.  The outsider will have to take a leap of faith, or a step in their imagination, to accept that they need OR to deal with their strategic problems.  Articles in a magazine may help to convince them, but first they have to read the article.  I hope that some of those outsiders have found OR Insight of interest and that some have realised that they need OR in their organisations or businesses.

In due course the OR Society is going to replace OR Insight.  Watch this space!



Friday, 13 December 2013

Multi-Criteria Decision Analysis (MCDA) in the Stone Age

From The Independent (UK daily newspaper) dated Wednesday 11th December 2013
Stone Age Brits were past masters at choosing the perfect ‘des res’, according to new research carried out by archaeologists.
Their investigations have revealed that, 300,000 years before the emergence of anatomically modern humans, prehistoric Britons were selecting their domestic real estate with tremendous care.
Nutritional and security considerations appear to have been the main criteria, according to the new research carried out by scholars at the University of Southampton and Queen's University, Belfast.
A survey of 25 major British and north-west French sites dating from 500,000 to 200,000 years ago has revealed that early humans – members of the now long-extinct species Homo heidelbergensis – predominantly chose to live on islands in the flood plains of major rivers. They avoided  forests and hills – and the upper and middle reaches of river systems,  and their estuaries.
It is the first ever detailed interdisciplinary investigation  into early humanity’s home location preferences. The degree to which they preferred to  choose just one specific type of location has surprised the archaeologists.
Full story is here.

The article continues to describe some of the common aspects of those village sites:
  • good water supply;
  • somewhere that cereals can be found;
  • convenient for hunting animals;
  • convenient for fishing;
  • safe from surprise attack.
The way that the story was written made both Tina and I think that the writer had chosen an anachronistic way of picturing the situation.  So, here is the scenario: the tribal leader of Homo heidelbergensis gathers members of the tribe around him (assuming male dominance) and asks them:
"We have five possible sites for our village.  Please will you help me choose the best one?  I need some answers to help me perform my MCDA.
  • On a scale of one to ten, how do you rate each site for water supply?
  • On a scale of one to ten, how do you rate each site for finding cereals?
  • On a scale of one to ten, how do you rate each site for hunting?
  • On a scale of one to ten, how do you rate each site for freedom from attack?
  • On a scale of one to ten, how do you rate this idea of MCDA?
  • Who taught you to count up to ten anyway?
  • On a scale of one to ten, how do you rate each site for  leaving evidence for 21st century archaeology?
  • Why am I using the word multi-criteria, when it is derived from a language that hasn't been invented yet?"
Somehow, I don't think that is what really happened in the Stone Age.  The tribes used a different problem solving approach.  They used heuristics.  A site was used, and if they liked it, they stayed there.  If they found somewhere better, they moved.  Gradually, they would settle in a site for long enough to leave evidence of their occupation for modern archaeologists to find.  The same process of adaptation of the location of a village site has happened over and over again over the millennia.  In terms of the age of the human race, the concept of having a fixed site is comparatively recent.  Our ancestors were nomadic.  Ful stop.

And, with a sideways look at another story from history, I started to think of the early colonists of North America.  The Pilgrim Fathers survived, partly through the help of local Indians, who had - presumably - used the same heuristics to find where to live through their knowledge of the territory around Plymouth, Massachusetts.  Other early colonists may not have survived long enough to acquire that essential local knowledge.

It also makes one think about those works of science fiction and fantasy where colonists arrive at a new planet and start to settle in.  For the sake of the plot lines, everything usually goes well from the outset.  But those colonists have come from a society which has lost the expertise about site selection that Homo heidelbergensis possessed.  That would be a gap in their knowledge, one that heidelbergensis took generations to refine.

Nonetheless, I still like the idea of MCDA hundreds of thousands of years ago.

Thursday, 5 December 2013

Harriet Vane and food statistics

In Dorothy L Sayers' book, Gaudy Night, Harriet Vane, the heroine, meets the cook at Christ Church after being knocked over by the nephew of her husband-to-be, Lord Peter Wimsey. 
The Christ Church cook was well pleased to produce meringues from the ancient and famous College oven; and when Harriet had duly admired the vast fireplace with its shining spits and heard statistics of the number of joints roasted and the quantity of food consumed in term-time, she followed her guide out into the quadrangle again ...
The kitchen at Christ Church, as it used to be

In several of Dorothy Sayers' books, there are references to statistics; the writer worked for a time in the advertising industry, where she would have been confronted with measures of effectiveness.  Another book, Five Red Herrings, needed detailed research about train times and road distances.  In this respect, she is much more careful about her facts than was Sir Arthur Conan Doyle, who has given generations of readers questions about which train Sherlock Holmes would have caught on different occasions.
Oxford and Cambridge college kitchens still turn out huge numbers of meals, but today the statistics would have been about avoiding food wastage.  St John's College, Cambridge (on page 16 of the news)  has just won national awards for its sustainable food policy, and records that it has very little food wastage, less than 5%.  Harriet would be impressed.
It made me wonder how different businesses in the food industry could reduce their wastage rates; there's scope for some interesting O.R. studies here.