An orthogonal hyperprismatic grid whose all nodes are accompanied by the given function Values Call not be generally constructed due to the random nature of the given function data. This prevents the reduction Of the single multivariate interpolation to More than One univariate or bivariate interpolations even approximately. It is generally quite difficult to determine an analytical structure for the target function in these problems. Lumping HDMR method is an indexing based High Dimensional Model Representation (HDMR) algorithm used to reconstruct these types Of multivariate data by imposing an Indexing scheme to obtain an orthogonal geometry for the given problem. By this way, the training Of the given data call be accomplished. The next problem, is to determine I reverse algorithm for the testing data. This work is about it new algorithm to find the correct coordinate of the given testing data in the orthogonal geometry obtained through Lumping HDMR.