Identifying Productivity Drivers by Modeling Work Units Using Partial Data

We describe a new algorithm for estimating a model for an independent variable which is not directly observed, but which represents one set of marginal totals of a positive sparse two-way table whose other margin and zero pattern are known. The application which inspired the development of this algorithm arises in software engineering. It is desired to know what factors affect the effort required for a developer to make a change to the software: for instance, to identify difficult areas of the code, measure changes in the code difficulty through time, and evaluate the effectiveness of development tools. Unfortunately, measurements of effort for changes are not available in historical data. We model change effort using a developer's total monthly effort and information about which changes she worked on in each month. We produce measures of uncertainty in the estimated model coefficients using the jackknife by omitting one developer at a time. We illustrate a few specific applications of our tool, and present a simulation study which speaks well of the reliability of the results our algorithm produces. In short, this algorithm allows analysts to identify if and how much impact a promising software engineering tool or practice has on coding effort.