DATA MINING TECHNIQUES TO PREDICT INFORMATION SYSTEMS MAINTAINABILITY

摘要

This paper presents an approach, based on Data Mining techniques, to predict the maintainability of class diagrams done using the Unified Modelling Language (UML), which has great importance on the final quality of object -oriented information systems (OOIS). The prediction model is built from metrics values obtained at the early phases of OOIS life-cycle. We will start with Fuzzy prototypical Knowledge Discovery process (FPKD) for finding Fuzzy Deformable Prototypes of class diagram maintainability, and later we will predict a real case class diagram maintainability deforming the similar prototypes using the degree of compatibility with them. The data used for prediction was obtained through a controlled experiment.