Performance-Enhanced Hybrid Memetic Framework for Effective Coverage-Based Test Case Optimization

Abstract

The rapid advancement of computer and database technologies has increased the importance of feature selection. The filter methods or wrapper methods may be used for feature selection. The filter method is computationally cheap but has a risk of selecting subsets of features that may not match the chosen induction algorithm, whereas the wrapper method engages the induction algorithm to evaluate the feature subsets and is computationally more intensive and has better prediction accuracy than filter method. Memetic algorithm, a family of metaheuristics coined by R. Dawkins from the term meme, denotes an equivalent to the gene to maintain a population pool comprising several solutions simultaneously for the problem. In this research work, the authors propose a hybrid wrapper–filter feature selection algorithm (WFFSA) using a memetic framework for effective coverage-based test case optimization. The test cases are optimized using a hybrid memetic framework method by combining the call stack and memetic algorithm. A call stack is a sequence of active calls associated with each thread in a stack-based architecture. Methods are pushed onto the stack when they are called and are popped when they return or when an exception is thrown. In this research, each of these solutions is termed individually, and an empirical experiment has also been conducted to demonstrate that the hybrid method improves the performance and results in reduction of the cost factor.