search menu icon-carat-right cmu-wordmark

Causal Models for Software Cost Prediction and Control

October 2019 Presentation
Michael D. Konrad, Robert W. Stoddard, William Nichols, David Zubrow

This presentation describes work toward establishing causation with observational data, which remains a vital need and a key technical challenge.

Abstract

Cost estimation inaccuracy continues to be cited as a dominant factor in DoD cost overruns. Research has shown causal models are superior to traditional statistical models because, by identifying truly causal factors, proactive control of project and task outcomes is possible. In this work, we expect to develop causal models, including structural equation models (SEMs) that provide a basis for (1) calculating the effort, schedule, and quality results of software projects under different scenarios (e.g., traditional vs. agile) and (2) estimating the results of interventions applied to a project in response to a change in requirements (e.g., a change in mission) or to help bring it back on track toward achieving cost, schedule, and technical requirements.