Resumen
This paper gives a comparative study of the K-means algorithm and the mixture model (MM) method for clustering normal data. The EM algorithm is used to compute the maximum likelihood estimators (MLEs) of the parameters of the MM model. These parameters include mixing proportions, which may be thought of as the prior probabilities of different clusters; the maximum posterior (Bayes) rule is used for clustering. Hence, asymptotically the MM method approaches the Bayes rule for known parameters, which is optimal in terms of minimizing the expected misclassification rate (EMCR).

Resumen
This paper addresses the problem of power and sample size calculation for a stepwise multiple test procedure (SD2PC) proposed in Tamhane et al. [2001. Multiple test procedures for identifying the maximum safe dose. J. Amer. Statist. Assoc. 96, 835–843] to identify the maximum safe dose of a compound. A general expression for the power of this procedure is derived. It is used to find the minimum overall power and minimum power under the constraint that the dose response function is bounded from below by a linear response function. It is shown that the two minima are attained under step and linear response functions, respectively. The sample sizes necessary on the zero dose control and each of the positive doses to guarantee a specified power requirement are calculated under these two least favorable configurations. A technique involving a continuous approximation to the sample sizes is used to reduce the number of quantities that need to be tabled, and to derive the asymptotically optimal allocation of the total sample size between the zero dose and the positive doses. An example is given to illustrate use of the tables. Extensions of the basic formulation are noted.