Chi-Square Goodness-of-Fit Tests for Strong-Mixing Stationary Processes.

摘要

An attempt has been made in this report to investigate whether appropriate tests can be formulated to test the adequacy of various models about the univariate distribution of a strictly stationary process. Specifically,it is assumed that the strictly stationary process is strong-mixing,that the data derived from such a process is grouped and that the standard chi-squared tests based on the deviations of the observed from the expected frequencies for these groups are used. Asymptotic sampling properties of these tests for various types of simple and composite hypotheses about this univariate distribution have then been derived.