Can Taguchi Method be applied to Improve Software Quality ? A Case Study
Abstract
Taguchi method was formulated by Dr. Genichi Taguchi for the purpose of improving quality and hence the productivity of industrial products, ref. Burgam (1965), Moen et al. (1992), Peace (1992), Roy (2010). Since then several US corporations have commenced utilising the method for the above purpose. Basic approach of the Taguchi method is to use a think tank of persons employed in the production to identify the components to be called factors and the levels of each of its use. Different combination of factors and uses are called trials. The resulting production level for each trial is measured and the combination that provides the optimal value is statistically determined. Since software has parallels with industrial or service production, Bagehi et al. (1992), Ravella (2008), Kanchana et al. (1999), had reported on some attempts that have been made to apply Taguchi method to quality improvement of software systems. Instead of using Lf trials; (L - no. levels, f – no. factors for each), Taguchi method uses a lesser number of trials using a concept of orthogonal arrays. A statistical technique close to principal component analysis is then used to determine the trial pattern that is expected to yield the highest quality ranking Taguchi method relies heavily on the use of a think tank of around 10 – 15 members. The main task assigned to its members is to give a possible score, for instance from 0 (very bad) to 10 (excellent) for each trial and the average is taken as the score. An interesting feature of the method is that the optimal trial pattern may sometimes not be one of the orthogonal arrays mentioned. Objective of this research is to investigate the possibility of using the Taguchi method to determine how best the resources, viz., factors and the level of each that should be used to obtain maximum possible quality level of the finished product.