New Zealand Statistical Association Newsletter 61
March 2005
Mathematics-in-Industry Study Group
MISG-05 Needed Statistical Input
The ANZIAM Mathematics-in-Industry Study Group held in Massey University,
Auckland in January 2005 (see
http://misg2005.massey.ac.nz/) was presented with two problems (out of a
total of seven) which had a particularly strong need of expert statistical
input. These are described below:
1. Development of empirical relationships for metallurgical design of hot-rolled steel products. (New Zealand Steel Ltd, Glenbrook).
See http://misg2005.massey.ac.nz/problems/problem2.html“An empirical model is required which relates the mechanical properties of hot-rolled coil products produced by New Zealand Steel to product and processing variables such as chemistry, and rolling and coiling temperature.”
2. Factors associated with trends in bare ground in high country. (Environment Canterbury, Christchurch).
See http://misg2005.massey.ac.nz/problems/problem4.html“Environment Canterbury anticipates obtaining a mathematical model that identifies and describes the main effects and interactions of the factors influencing short and long term ground cover trends in the high country. Such a model would be invaluable for identifying land management options that could be applied to help ensure soil conservation in the Canterbury high country tussock grassland ecosystems.”
This served to underline the experience many of us have - that the first encounter many industries have with the quantitative sciences is: “I have all this data, what does it mean?” In both cases the organisations had put aside the immediate goal of finding a more mechanistic / modelling-based algorithm. In one case (NZ Steel) subsequent effort was directed a little towards finding a more physically based model, but this was not their major goal.
MISG was fortunate that it had attracted senior well-known statisticians who had independently recognised this need for these problems (Associate-Professors Ken Russell (University of Wollongong) and David Scott (University of Auckland) among others, and they provided key leadership in the analysis and resolution of these problems. We acknowledge their leadership in this respect and suggest that future MISGs may have similar such requirements, and invite the statistical community to participate (It’s free and interesting!) [6 members of NZSA were among the 125+ international participants at MISG2005 - Ed.] For further details see http://misg2005.massey.ac.nz. The next ANZIAM MISG is being held in the same venue early in 2006. See the notice below.
Graeme Wake, Centre for Mathematics-in-Industry,
Massey University, Auckland.
Mathematics-in-Industry Study Group 2006
30th January -3rd February 2006
http://misg2006.massey.ac.nz
It immediately precedes the Applied Mathematics Conference ANZIAM2006, in Mansfield, Victoria.
Recordings from the ANZIAM MISG 2005
The Centre for Mathematics in Industry at Massey University, Auckland, NZ has available a set of DVDs of the formal presentations of the Monday and Friday sessions of MISG2005. We can place bulk orders for these at just $NZ 24 + GST (= $NZ 27) for each set. They include the power point and video presentations of “Problems” and “Solutions” of the seven problems packaged together, and the opening ceremony. These were produced professionally. Four discs with: 2 Problems & Solutions on each disc.. They are excellent for promotional purposes. Orders to:
Professor Graeme Wake,
Centre for Mathematics in Industry,
Massey University at Albany,
P.B 102904, North Shore MC,
Auckland, New Zealand.
Please include cheque (made out to Massey University) or Visa/MasterCard details with each order.
E-mail g.c.wake@massey.ac.nz.
Reflections on MISG-05
The 2005 Mathematics-in-Industry Study Group took place at the end of January at the Albany Campus of Massey University. You probably heard about it. It was opened by Dr Michael Cullen, Deputy PM of NZ, who suggested that it was time for New Zealand to put behind it the traditional reliance on inventors with number 8 fencing wire, and to bring a scientific and mathematical approach to bear. (Dr Cullen is not responsible for my paraphrasing of his reported remarks.) Naturally, the press ‘beat up’ these comments.
MISGs have occurred annually in Australia or New Zealand for a good number of years. Companies can, for a modest fee, propose a problem for solution, and enjoy the input of some very able people who (hopefully!) either solve the problem or advance its solution by several giant steps. In the past, statisticians have not usually attended MISGs and most problems have been viewed as requiring Applied Mathematics for their solution. It is unclear whether the absence of statisticians caused, or was caused by, this view!
This year I was one of a small group of statisticians who attended MISG 2005. It was instantly apparent from reading the list of seven problems that at least two of them were heavily statistical in nature: they essentially required the fitting of multiple regression models in which several of the explanatory variables were categorical. The applied mathematicians were painfully aware of their inadequacies in this regard, and the statisticians were made very welcome.
On the positive side, I enjoyed the camaraderie of working with other researchers, both statistical and mathematical, and was highly impressed by the extreme professionalism of my group’s facilitator (take a bow, Heather North) and industry contact (Jeromy Cuff). It was also great to catch up with some mathematicians whom I’d known in Wellington over 20 years ago. However, it was depressing to realise how little Statistics most applied mathematicians knew (after all, a ‘one-way ANOVA’ is pretty basic stuff), to wonder how they might have got on without statistical help, and also to wonder whether there might have been unrecognised situations like this at past MISGs. I was also conscious that there was very little by way of computing support available, and my group would have been completely lost without the laptops, and clever work, of Barry McDonald and Ray Hoare.
Of the seven problems posed, I feel that at least partial solutions were obtained for six of them, and perhaps the seventh does not have a solution. New Zealand industry, and the Canterbury High Plains, will be the better for our efforts. MISG has succeeded again.
Elsewhere in this Newsletter, there is an invitation from Graeme Wake for statisticians to attend the next MISG. I encourage you to give this serious thought. Graeme is an energetic and enthusiastic organiser who recognises the need for statisticians (even Australian statisticians), and I am sure that you will be made very welcome. We need to increase the profile of Statistics amongst our mathematical colleagues and in industry, and this is a very good way to do so. I suggest that you check the program to be sure that there are some problems with a statistical bent before registering, and take a laptop loaded with your favourite statistical software if you can.
Lastly, can you justify asking your employer for support to attend MISG 2006? If you’re a PhD student needing real experience, or a beginning statistician who would like to dip your toe in the consulting water in a supportive environment, the answer is definitely ‘yes’. If you’re an experienced consultant who has already fitted lots of regression models to messy data, then MISG 2005 needed you, and MISG 2006 probably will, too, but I’m not sure that there’s an easy way to justify your attendance. Altruism, perhaps?
Ken Russell