عنوان مقاله [English]
For many years, analysis of real systems has attracted much attention. Such systems are hard to describe because of their complex behavior and their huge number of parameters and mutual effects. This issue has made analysis methods of a multivariable system develop rapidly. Multivariate analysis consists of a collection of methods that can be used when several measurements are made of each individual or object in one or more samples. In practice, multivariate data sets are common, although they are not always analyzed in the same way. However, the exclusive use of univariate procedures with such data is no longer acceptable, given the availability of multivariate techniques and inexpensive computing power. The Mahalanobis-Taguchi system is such a novel method. Nowadays, rapid development of technology has made it possible for organizations to gather large amounts of data for analyzing processes. But, on the other hand, an appropriate approach for dealing with these huge amounts is also required in modern commerce. The main purpose of this paper is to use Mahalanobis-Taguchi systems for effective selection after reorganizing and omitting non-meaning data. This method is for multivariable systems and consists of two parts: The first part deals with a useful variable selection for complexity reduction and the second part contains recognition and prediction processes and identification of abnormal groups. The conventional method uses orthogonal Taguchi arrays for variable selection. In fact, the main novelty of this paper is the use of concepts such as mis-classification and Integer Programming. The solution in our method is based on a creative algorithm, which is also accurate enough. It will be shown that this method is faster than former methods, and that its performance is generally better than previous methods.