Rough Sets were introduced by Zdzislaw Pawlak on his book "Rough Sets: Theoretical Aspects of Reasoning About Data". Rough Sets provide a formal method to approximate crisp sets when the set-element belonging relationship is either known or undetermined. This enables the use of Rough Sets for reasoning about incomplete or contradictory knowledge. A decision table is a prescription of the decisions to make given some conditions. Such decision tables can be reduced without losing prescription ability. This package provides the classes and methods for knowledge reduction from decision tables as presented in the chapter 7 of the aforementioned book. This package provides functions for calculating the both the discernibility matrix and the essential parts of decision tables.
This package allows users to reduce knowledge by simplifying rules in decision tables according to Rough Set theory. It is not a complete implementation of Rough Set theory; it just has the minimum functions to simplify decision tables.
The "RoughSetKnowledgeReduction" package pretends to be an R implementation of the section 6.3 "Simplification of Decision Tables" of the book written by Pawlak (Rough Sets: Theoretical Aspects of Reasoning About Data). In the context of artificial intelligence, Rough Set theory is considered a solution alternative to the classificatory problem, allowing to discard superfluous or irrelevant information to focus in the most determinant conditions for taking a decision. Rough Set theory is known for being able to deal with contradictory or even incomplete information, making no assumptions about the internal structure of the data. Rough Sets theory is unable to deal with continuous variables which is a clear disadvantage. For more details about Rough Set theory consult Pawlak's book.
For more details, download the PDF at https://github.com/albhasan/RoughSetKnowledgeReduction/blob/master/vignettes/howto.pdf?raw=true