Evolutionary Computing based an Efficient and Cost Effective Software Defect Prediction System
The earlier defect prediction and fault removal can play a vital role in ensuring software reliability and quality of service. In this paper Hybrid Evolutionary computing based Neural Network (HENN) based software defect prediction model has been developed. For HENN an adaptive genetic algorithm (A-GA) has been developed that alleviates the key existing limitations like local minima and convergence. Furthermore, the implementation of A-GA enables adaptive crossover and mutation probability selection that strengthens computational efficiency of our proposed system. The proposed HENN algorithm has been used for adaptive weight estimation and learning optimization in ANN for defect prediction. In addition, a novel defect prediction and fault removal cost estimation model has been derived to evaluate the cost effectiveness of the proposed system. The simulation results obtained for PROMISE and NASA MDP datasets exhibit the proposed model outperforms Levenberg Marquardt based ANN system (LM-ANN) and other systems as well. And also cost analysis exhibits that the proposed HENN model is approximate 21.66% cost effective as compared to LM-ANN.
