Weight and Cluster Based Test case Prioritization Technique

Abstract

Software testing has a significant importance to achieve maximum quality to satisfy the customers and concerned stakeholders. A test case is designed to perform set of actions with intend of finding errors and verify some functions and features of an application. During design process, a huge number of test cases produced, some of them are of little or no use, which can be ignore or postponed, when there is budget and time constraints, or a need to decide which test cases to execute first and which to last. However, in black box testing, test cases are prioritized manually during planning phase and companies mostly experience schedule limitations, in that case, effective testing costs them badly. Test case prioritization's main purpose is to effectively use time and budget to execute highest priority test cases first with customer's satisfaction. To achieve this goal, we proposed a technique in which we use a customer assigned weight abstracted from business requirements to keep the customer's preference first, based on that three main clusters formed. Then we calculate proposed cost and time percentage for each test case using function points and complexity measure, with in each cluster. Based on that, clusters further classified in to High, Medium and Low priorities clusters by K-Medoids algorithm. In our approach, test cases finally classified in to clusters and sub clusters based on the priority of the both stakeholders. Our approach shows 79.174% accuracy as compared to the actual data. To achieve maximum efficiency, considering user's satisfaction, this method of mining test cases will be helpful in terms of saving time and cost.