A NOVEL APPROACH IN THE INSPECTION PROCESS WHICH MEASURES THE QUALITY OF INSPECTION THROUGH THE INTRODUCTION OF A NEW PROCESS METRIC CALLED AS 'DEPTH OF INSPECTION' (DI) FOR EFFECTIVE DEFECT MANAGEMENT IN SOFTWARE DEVELOPMENT

摘要

Conduction of inspection during software development life cycle plays a vital role in effective defect management and thereby enables one to develop a quality software product. However, quality must be a quantifiable unit. A software quality metric that can measure the quality of software inspection did not exist so far. Therefore, we have introduced a new approach of inspection which consists of two stages. Stage-1 consists of inspection activity performed by a minimum of three inspectors and stage-2 will be the final inspection. The resultant of both stages is an Inspection Quality Analysis report (IQA). IQA contains information which includes Depth of Inspection (DI) value, Inspection Performance Metric (IPM) value etc. DI is a process metric which measures the depth in which inspection has occurred. DI enables the manager within the software community to identify and compare the level of inspection performed in various projects. DI further facilitates one to have a deep visibility of the process and helps to control the developmental cost. Thus, introduction of DI in software industry can yield valuable information of a company in relation to the inspection process and effectively reduce further burdens that remain ahead. DI implementation as a process metric can be realized in two phases. In the first phase DI can be calculated for a particular project based on the shop floor defect count. The second phase of DI enables the software company to predict depth of inspection using a mathematical modelling scheme. DI (Y) variable depends on four mutually independent variables (x1 to x4) and on four process coefficients ( β1, ( βn). The four variables are the four process parameters which are inspection time, inspection preparation time, number of inspectors and experience level of inspectors. The second phase of DI prediction requires a minimum of five projects to evaluate the process coefficients using pivot solutions like least square technique. With stabilized process coefficients and chosen process parameter value, it is now possible to predict DI for any project (Pi) based on a mathematical model where the properties of the company are evaluated on a regression basis. Our invention shows the modified approach of software inspection where the process is measured using DI metric in software industry for the benefit of Software Engineering world.