Statistické neparametrické metody založené na iteracích

Abstract

For many of the statistical tests it is necessary to assume, that the population is normally distributed. These methods are called parametric methods. Tests in which such a strong assumption as normality doesn't need to be made about the population distribution are called nonparanetric tests. Some of less known nonparametric methods are based on iteration theory. Iterationis a sequence of data that exhibit the same characteristic (either 0 or 1) ; the sequence is preceded and followed bz different data or no data at all. Iteration length is determinated bz number of data in the iteration. Number of iteration shouldn't be neither too great, nor too small. Generally on base of the number of iteration of values 0 and 1 we can estimate, whether the ordering of the values 0- and 1 in sequence is random. Critical value salfa is searched in the table. The aim of this contribution is to find probability distribution of number of the iteration. In next part there're presented three of long line of possible employment of iteration theory. a) On base of presented theory we can test hypothesis , that two independent sets of data come from population with a common distribution. b) Test for randomness of sample data. This test is based on the order, in which the data occur, not on the frequency of the data. c) Sign test.