Degree classification and faults classification

well as make necessary validations on their hypothesis by using real data. In classification, as mentioned in Chap. , one purpose of rule representation is to present the knowledge in different ways in accordance with specific commercial requirements. From this point of view, the networked representation can effectively reflect the importance of input attributes and provide a ranking of the attributes according to their importance. In practice, each input attribute can be seen as an impact factor to which a decision outcome is subject. And the ranking of attributes can help domain experts identify which ones are more important, less important or irrelevant to the decision outcome.

This book introduces some novel methods and techniques in rule based classifi-cation and ensemble learning in Chaps. , , and namely information entropy based rule generation (IEBRG), Jmid-pruning, rule based networks, collaborative and competitive decision rules and hybrid ensemble rule based classification.