ANALIZA WYBRANYCH METOD ESTYMACJI PARAMETRÓW GRANICZNYCH ROZKŁADÓW STATYSTYKI MAKSIMUM
ANALYSIS OF CHOSEN ESTIMATION METHODS OF MAXIMUM STATISTIC LIMIT DISTRIBUTION PARAMETERS

Summary/Abstract: The limiting distribution of maximum statistic determined on the basis of a random sample is one of the following distributions: Gumbel, Fréchet or Weibull. If we have information about the distribution class of the analyzed variable we use limit theorems about maximum distribution, otherwise we must apply appropriate statistical tests based on the order statistics. We can use different methods to estimate the parameter of maximum distribution, in particular the maximum likelihood method, the method of moments and methods based on quantiles. The paper presents the results of analysis of mean squared errors of Gumbel distribution parameters estimators obtained by the methods of moments, the quantile method and the quantile least squares method with a truncated number of quantiles. Received results allow to draw conclusions on the regarded estimators properties, specifically the efficiency of the chosen estimation methods.

• Details
• Contents