Comparative Analysis of Software Components Reusability Level using GFS and ANFIS Soft-Computing Techniques

Abstract

The quest to develop software of great quality with timely delivery and
tested components gave birth to reuse. Component reusability entails
the use (re-use) of existing artefacts to improve the quality and
functionalities of software. Many approaches have been used by
different researchers and applied to different metrics to assess software
component reusability level. In addition to the common quality factors
used by many authors, such as customisability, interface complexity,
portability and understandability, this study introduces and justifies
stability, in the context of volatility as a factor that determines the
reusability of software components. Sixty-nine software components
were collected from third party software vendors and data extracted
from their features were used to compute the metric values of the five
(5) selected quality factors. Genetic-Fuzzy System (GFS) was used to
predict the level of the components’ reusability. The performance of the
GFS was compared with that of Adaptive Neuro-Fuzzy Inference
System (ANFIS) approach using their corresponding average RMSE
(Root Mean Square Error), in order to ascertain the level of accuracy
of the prediction. The results of the findings showed that, GFS with an
RMSE of 0.0019 provides better reusability prediction accuracy
compare to ANFIS with an RMSE of 0.1480.