Investigating the Accuracy of Test Code Size Prediction using Use Case Metrics and Machine Learning Algorithms
Mourad Badri, Linda N. Badri, William Flageol, Fadel Touré · 2017
Software testing plays a crucial role in software quality assurance. It is, however, a time and resource consuming process. It is, therefore, important to predict as soon as possible the effort required to test software, so that activities can be planned and resources can be optimally allocated. Test code size, in terms of Test Lines Of Code (TLOC), is an important testing effort indicator used in many empirical studies. In this paper, we investigate empirically the early prediction of TLOC for object-oriented software using use case metrics. We used different machine learning algorithms (linear regression, k-NN, Naïve Bayes, C4.5, Random Forest, and Multilayer Perceptron) to build the prediction models. We performed an empirical study using data collected from five Java projects. The use case metrics have been compared to the well-known Use Case Points (UCP) method. Results show that the use case metrics-based approach gives a more accurate prediction of TLOC than the UCP method.