Using Structured Text Source Code Metrics and Artificial Neural Networks to Predict Change Proneness at Code Tab and Program Organization Level

Abstract

Structured Text (ST) is a high-level text-based programming language which is part of the IEC 61131-3 standard. ST is widely used in the domain of industrial automation engineering to create Programmable Logic Controller (PLC) programs. ST is a Domain Specific Language (DSL) which is specialized to the Automation Engineering (AE) application domain. ST has specialized features and programming constructs which are different than general purpose programming languages. We define, develop a tool and compute 10 source code metrics and their correlation with each-other at the Code Tab (CT) and Program Organization Unit (POU) level for two real-world industrial projects at a leading automation engineering company. We study the correlation between the 10 ST source code metrics and their relationship with change proneness at the CT and POU level by creating experimental dataset consisting of different versions of the system. We build predictive models using Artificial Neural Network (ANN) based techniques to predict change proneness of the software. We conduct a series of experiments using various training algorithms and measure the performance of our approach using accuracy and F-measure metrics. We also apply two feature selection techniques to select optimal features aiming to improve the overall accuracy of the classifier.