Authors : Safa Saad A. AL-Murieb and Fryal Jassim Abd Al-Razaq
Abstract: This strudy presents the use of the regression capability of the TreeNet regression tool to predict software class testing effort using a dataset obtained from Apache Ant 1.7.0. A total of twelve models with different loss regression criterion, were built with each set of predictor variables to predict the code lines (DLOC) and the Test Cases Number (NTC) of a test class. Six of the models were built using all the variables and the remaining six with subsets of the variables that were found important to their respective models. The subsets of variables were selected using a stepwise regression tool. The comparison between the TreeNet regression models results and Multi-Linear Regression (MLR) showed that the TreeNet regression models are better and further confirmed the prediction capability of the TreeNet regression tool.
Safa Saad A. AL-Murieb and Fryal Jassim Abd Al-Razaq, 2018. TreeNet Model Based Prediction of Testing Effort Class. Journal of Engineering and Applied Sciences, 13: 4687-4693.