|There are several evaluation methods for parameter estimation. For example, the mean squared error (MSE), the Euclidean distance, the Mahalanobis distance and etc. But, when statistical methods have two or more parameter, it is difficult to evaluate whether they are good or not. In this paper we focus on defining the generalized bias (gBias) and the generalized mean squared error (gMSE) related to the generalized variance (gVar) applying to MSE concept to evaluate the parameter estimates. Moreover we will prove this concept beyond the three-dimensional, that is, hyper dimensional. Also we will check the applications of these terms. We will consider the robust estimates such as MAD, Hodges Lehmann estimator compared to other evaluation methods. We will use nuclear area data and find efficient method to predict Weibull distribution in some condition.