Al has now provided some effective techniques for formalization of knowledge about goals and actions. These techniques has opened new research areas, Statistical Expert System, to statisticians. This statistical software is more intelligent - 'user-protective' and 'user-friendly-computer program using the formalized knowledge of professional statisticians in three statistical strategy levels : (i) problem formalization, (ii) statistical technique selection, and (iii) technique application. This thesis provides guidelines for a systematic approach to statistical expert system, based upon existing experimental systems.
Up to now, building consultation system in the technique application level is best explored. However, the experimental design or problem formalization is the most difficult and important level in real consultation. In this thesis, a statistical strategy for planning of experiments to be used in the expert system is developed for statisticians and others with a minimal knowledge of statistics. The suggested strategy follows 'the scientific consulting process' for planning of experiments. For non-statisticians the package to be developed using this strategy shall contain guidance, explanation, validation and implementation facilaties, which aims to substitute for a statistical consultant. The strategy for planning of experiments consists of 3 stages : (i) finding treatment structure, (ii) finding block structure, and (iii) randomly assigning treatments to the experimental units. The scope of this thesis mainly includes the generalized factorial experiments.
