ABSTRACT 
In Part I, the general procedures of the conventional method of correlation or 
spectrum analysis of a random process (nonparametric method) are reviewed, stressing 
the statistical reliability of the results. A few suggestions for improving the coherencies 
The model-fitting technique supported by AIC criteria is introduced, with the examples 
of application to the seakeeping data. In Part II, the statistical treatments of nonlinearities 
in random process analysis are summarized and reviewed. Conclusions are given, and 
future work is proposed. 
