With the growing demand for software-driven hardware and software driven processes, the demand for building software applications are also increasing. From the applications consumer or the application owner perspective, the functionality may be different for each application, but, the applications are built using simple building blocks. Analysing a code or code segments, to be justified for code reusability or code segment replaceability, is a highly challenging task. A good number of parallel researches are carried out to identify a method to accurately detect software code segment equivalence for reuse. Even so, the attempts have failed to achieve higher accuracy due to highly diverse method signatures and the class hierarchies for the software applications. Hence, this paper attempts to achieve a novel framework to predict the software code segment reusability using four different algorithms in a framework for analysing time series-based software metric data, which are extracted by various researchers as a standard dataset. The framework is built to analyse the time-series-based metric data as the fundamental nature of the enterprise applications is to keep growing higher and higher by introducing multiple new software components. To improve the predictability and identification of pattern orientation of the software metric data, by applying the proposed pre-processing technique for identifying and replacing missing values to reduce the software component cyclic dependencies for finding the suitable threshold for comparing the software components. After applying pre-processing techniques, for finding the best equivalence ranking, for the software components.
Volume 11 | Issue 7
Pages: 236-246