A new software maintainability evaluation model based on multiple classifiers combination
Fei Ye, Xiaodong Zhu, Yigang Wang · 2013
A Software Maintainability Evaluation Model based on Multiple Classifiers Combination (SMEM-MCC) is proposed, which is a software metrics based evaluation method. The model includes three parts: attributes selection, model training and model interpretation. Attributes are selected using a classifying selection method based on Genetic Algorithm (GA). The sub-classifiers of the integrated model are assembled by a BP NN. A rule extracting algorithm based on decision tree is used to interpret the results of the integrated model. A Software maintainability experiments is conducted, and a dataset which includes 300 software's class design metrics is achieved. The SMEM-MCC is trained and evaluated based on the dataset. The predication results show that the model proposed in this paper work better than any other single classifier, such as BPNN, SMO or decision tree.