Regression Testing Approach for Large-Scale Systems

Regression testing is an important and expensive activity that is undertaken every time a program is modified to ensure that the changes do not introduce new bugs into previously validated code. Instead of re-running all test cases, different approaches were studied to solve regression testing problems. Data mining techniques are introduced to solve regression testing problems with large-scale systems containing huge sets of test cases, as different data mining techniques were studied to group test cases with similar features. Dealing with groups of test cases instead of each test case separately helped to solve regression testing scalability issues. In this paper, we propose a new methodology for regression testing of large-scale systems using data mining techniques to prioritize and select test cases based on their coverage criteria and fault history.

I. Introduction

Testing is a vital activity in software engineering that controls the quality of software by detecting errors and verifying the completion of software functionality as defined by the requirements. Regression testing is a type of software testing concerned with testing changes done to a software to ensure that any modifications are done successfully without imposing other modules to produce unexpected results. It is typically done by re-running all the previous test cases which is an expensive and time-consuming activity that increases by increasing the size and complexity of software. Different techniques have been studied to solve regression testing problems; (i) test cases reduction, which permanently eliminates test cases from the test suite, (ii) test cases prioritization, which orders the test cases by certain measures and (iii) test cases selection which seeks to select test cases that are relevant to some set of recent changes [1], [2], [3].