Selecting a Defect Prediction Model for Maintenance Resource Planning and Software Insurance

Date of Original Version



Conference Proceeding

Rights Management

Can archive pre-print and post-print or publisher's version/PDF

Abstract or Description

Better post-release defect prediction models could lead to better maintenance resource allocation and potentially a software insurance system. We examine a special class of software systems and analyze the ability of currently-available defect prediction models to estimate user-reported defects for this class of software, widely-used and multi-release commercial software systems. We survey currently available models and analyze their applicability to an example system. We identify the ways in which current models fall short of addressing the needs for maintenance effort planning and software insurance.


This research was supported by the National Science Foundation under Grand CCR-0086003, by the Carnegie Mellon Sloan Software Center, by the High Dependability Computing Program from NASA Ames cooperative agreement NCC-2-1298. The authors would like to thank Peter Santhanam and Bonnie Ray of IBM Research for their contribution to this work. The authors would like to thank Audris Mockus for his insights, and Vahe Poladian for his valuable input.


Published In

Proceedings of the Fifth Workshop on Economics-Driven Software Research, IEEE Computer Society, 32-37.