An approach to fuzzy granule-based hierarchical polynomial networks for empirical data modeling in software engineering

  • Authors:

  • Byoung-Jun Park;Witold Pedrycz;Sung-Kwun Oh

  • Affiliations:

  • The Technology Research Institute, GM Tech Co., Ltd, Hwaseong-si, Gyeonggi-do, South Korea;Department of Electrical and Computer Engineering University of Alberta, Edmonton, Alta., Canada T6R 2G7 and Systems Science Institute, Polish Academy of Sciences, Poland;Department of Electrical Engineering University of Suwon, Hwaseong-si, Gyeonggi-do, South Korea

  • Venue:
  • Information and Software Technology
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Experimental software data capturing the essence of software projects (expressed e.g., in terms of their complexity and development time) have been a subject of intensive modeling. In this study, we introduce a new category of fuzzy granule-based hierarchical polynomial networks (FG-HPN) and discuss their comprehensive design methodology. The FG-HPN architecture benefits from the existence of highly synergistic linkages between fuzzy granules (referred here as granular information design phase of FG-HPN) and hierarchical polynomial networks (referred as network design phase of FG-HPN). We develop a rule-based fuzzy granules consisting of a number of ''if-then'' statements whose antecedents are formed in the input space and linked with the consequents (conclusion parts) formed in the output space. Hierarchical polynomial networks provide approximation of experimental data. In this framework, fuzzy granules contribute to the realization of the granular information design phase of the overall networks structure of the FG-HPN. The networks design phase is designed with the aid of genetically endowed hierarchical polynomial networks. The experiments reported in this study deal with well-known software data such as the NASA dataset concerning software cost estimation and the one describing software modules of the Medical Imaging System (MIS). In comparison with the previously discussed approaches, the proposed FG-HPN is more accurate and yield significant generalization abilities.