Over the years I have become increasingly concerned about the simplifying assumptions required in the standard application of the linear variate to data analysis. It seems to me that the simplifying assumptions (non-linearity, etc) do not hold up in the real world. I believe that one possible solution to the limitations imposed by these assumptions is the use of adaptive systems techniques, including such approaches as neural networks and evolutionary algorithms. My eventual aim is to apply these techniques to large scale marketing databases, an area which is starting to be known as knowledge discovery or data mining. To achieve that aim, I have been working on demonstrations of the technique on small scale datasets. The current dataset I am using is a questionnaire based survey of training needs in a professional organisation. Traditional approaches might use discriminant analysis or regression. What I am attempting to do is to develop an alternative approach to interpreting this dataset using an evolutionary algorithm. Such an approach requires the development of different ways of representing the data in a model, and different measures of the fitness of this representation. More generally I am interested in the impact and application of computational intelligence to marketing, and in the marketing of high-technology products.
Current 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996