Predicting Testing Effort Using Artificial Neural Network

The importance of software quality is becoming a motivating force for the development of techniques like Artificial Neural Network (ANN), which are being used for the design of prediction models of quality attributes. The purpose of this work is to examine the application of ANN for software quality prediction using Object-Oriented (OO) metrics. Testing effort has been predicted using ANN method and independent variables are OO metrics given by Chidamber and Kemerer. The public domain NASA data has been used to find the relationship between OO metrics and testing effort. The model has estimated testing effort within 35 percent of the actual effort in more than 72.54 percent of the classes, and with a MARE of 0.25. The results are quite interesting, however, more similar types of studies are required to be carried out with large data sets in order to establish the acceptability of the model. Keywords— Software quality, Measurement, Metrics, Artificial neural network, Coupling, Cohesion, Inheritance, Principal component analysis