LEAST-SQUARES ESTIMATION OF PARAMETERS FOR ASYMPTOTES OF EXTREMES,
Abstract
Consider a specified one of the largest order statistics of a sample of size n from a continuous population. Suppose that its large sample distribution is known to be of an identified one of the three parametric types (called first, second, and third asymptotes) that occur for the sample case. One problem considered is estimation of the parameters of this asymptote on the basis of a sample of size N from the same population, where N is substantially larger than n. The other problem is to estimate the parameters of a specified smallest order statistic of a sample of size n. Development of a nonparametric estimate of the asymptotic distribution for an extreme furnishes the basis for the estimation. The parameters are determined so that there is as close an agreement as possible, in a modified least-squares sense, between the asymptote and the nonparametric estimate. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 26, 1964
- Accession Number
- AD0433836
Entities
People
- John E. Waslh
Organizations
- System Development Corporation