Using In-Process Metrics to Predict Defect Density in Haskell Programs

In late-stage phases of development, action to correct defects can be cost prohibitive. Effective, efficient, and expressive measures of reliability during the development cycle could aid developers by providing early warning signs of where the system might require modification or further testing. To this end, this paper presents a method for estimating defect density in a system using a suite of internal metrics for Haskell programs. A feasibility study of this method was conducted by analyzing the source code of seven released versions of the Glasgow Haskell Compiler. Further studies are being conducted to refine the metric suite and to examine the potential of the method.