Construction of statistical analysis framework for predicting factors that affect code inspection

Competition in the software industry increases every passing day. Companies need to be kept at the highest level of customer satisfaction to be in this competition. The customer satisfaction is gained with the introduction of high-quality products. Error free software product is important criteria for software quality. Code inspection is an effective process that detects defects early and removes them from software products before testing. Measurement of effectiveness of the product increases quality of the software product and the process. Software quality metrics are used to measure the efficiency of the code inspection. The factors like preparation for the inspection, preparation time, number of inspectors, and the experience level of the inspectors affect the inspection productivity. These four are really important for the inspection and should be clearly defined at the beginning of the process. This thesis includes the metrics used for evaluation of inspection process, Depth of Inspection (DI), Inspector Performance Metric (IPM), and Defect Reduction Technique. A new system is developed by using these metrics and the estimation algorithms to have the best inspection time, preparation time, and inspector number and level of experience for the process at the beginning of the project. To compete the estimation operation linear regration, curve fitting and Gauss-Newton algorithm is being used.