Classification Rules for Multi-level Multivariate Data

In this article we study new classification rules for multiple m-variate observations over u-sites and over v-time points under the assumption of multivariate normality. We assume that the m-variate vector of observations has "jointly equicorrelated covariance" structure. The new classification rules are effective in discriminating individuals when the number of observations is very small and thus unable to estimate the unknown variance-covariance matrix. The classification rules are demonstrated by using a real world data set. Our results show that the performances of our new classification rules are superior to the traditional classification rule.