A Classification of Software Defect Prediction Models

Quality and security are major concerns in largescale software development. The early prediction of defective modules is becoming an important aspect in large-scale software systems to minimize resources spent (i.e., effort, time, etc.) to increase quality and security, and to reduce the overall cost of software production. This paper investigates different types of Software Defect Prediction Models and their evolution over time to assist researchers and practitioners in better understanding their maturity and help them in identifying the ones that best suite their needs since no model is able to perform well in any conditions. we have identified 11 different types of defect prediction models and we have investigated how those groups evolved over time. Among the 11 types of identified models, the Within-Project Defect Prediction Model is the most popular type of model from 1990 to 2019.