Building a knowledge base for correspondence analysis

This paper introduces a statistical strategy for Correspondence Analysis. A formal description of the choices, actions and decisions taken during data analysis is built. Rules and heuristics have been obtained from the application of this technique to real case studies. The strategy proposed checks suitability of certain types of data matrices for this analysis and also considers a guidance and interpretation of the application of this technique. Some algorithmic-like rules are presented and specific criteria are given for application. This strategy has been implemented in a statistical knowledge-based system prototype using hypertext and rule