| Popis: |
Let xi1, xi2,…, xin, for i=1, 2,…, k, denote independent ran- dome samples of size n each from k distinct populations, the probability density function (pdf) corresponding to the ith population being (l/bi)f[(x-ai)/bi], -∞ 0, where f(x) is a prototype pdf which defines the family of distributions common to the k populations. Let it further be assumed that the location and scale parameters, ai, bi, are unknown and are possibly different for the different populations. Suppose that we are able to assume that f is either one of the two specified families of pdf, f1 and f2. The problem of discrimination reviewed here is that of finding a decision rule which selects one of these two families as the true pdf. |
