Analytic Capabilities for Improved Software Program Management
November 2020 • White Paper
David Zubrow, Christopher Miller
This white paper describes an update to the SEI Quantifying Uncertainty in Early Lifecycle Cost Estimation approach.
Software Engineering Institute
The ever-increasing role of software in DoD systems is driving the necessity for software engineering organizations to adopt more robust analytic techniques to plan and manage software deliverables, quality, and overall software engineering capabilities. Today’s drive towards the Software Acquisition Pathways and DevSecOps makes these models highly valuable and relevant. This approach has a number of virtues but chief among them are (1) incorporating uncertainty into the inputs and assumptions associated with factors affecting project execution and success and (2) utilizing Monte Carlo methods to create an empirically derived distribution of predicted cost. Additionally, the approach incorporates causal learning to empirically discover relationships that can be used with confidence for planning, management and corrective action, and estimation. The combination of BBNs and causal learning together provides a practical way to learn from (1) all of the data generated by modern engineering environments and (2) to complement that with other sources as necessary. This yields a robust analytic capability that produces high quality and high confidence planning and management guidance for software development projects.