Technatomy has developed the System and Method for Probabilistic Modeling in a CMMI Level 4 Environment for Agile Software Development, a multi-dimensional tool suite to improve, manage and execute on the planning, performance and delivery of agile software development projects. The tools, methods and models apply CMMI Level 4 techniques to assess performance of processes on agile software development projects and involves applying structured, statistical measurement, analysis, planning and control to software development process that are inherently highly flexible and agile. In Agile software development, requirements and solutions evolve through the collaborative effort of self-organizing cross-functional teams of analysts, developers, testers and configuration managers. The Agile software development approach advocates adaptive planning, evolutionary development, early and continuous delivery, and flexibly responding to change. In order to achieve the probabilistic control of agile software development projects we have developed a suite of Agile development estimation and management capabilities including an Agile Development Process Performance Baseline (PPB), Agile Development Estimation Tool (ADET) Process Performance Models (PPM) and an Agile Project Performance (APP) tool. Collectively this suite of tools, along with their proprietary algorithms, deliver quantitative planning information and measurable, probabilistic forecasting of development performance with high confidence levels. The information used to prove the models is based on five years of software application projects spanning new development, sustainment, maintenance and DevOps.
The Agile Development PPB is the entry point to the process. This input is gathered from past performance project data that is computed utilizing tunable defined boundaries within dynamic algorithms to establish average (AVG), upper control limit (UCL), and lower control limit (LCL) boundaries and are used in measuring the health of the active project. The areas covered include: requirements, project velocity, developer velocity, testing and documentation. The PPB results feed into the ADET PPM, which ingests and processes the resulting data based on the Monte Carlo Simulation (MCS). Our ADET PPM implementation encompasses Probabilistic Results, Graphical Results, Sensitivity Analysis, Scenario Analysis, and Correlation of Inputs. The ADET PPM performs risk analysis and builds dynamic models of possible results by substituting a range of values—a probability distribution—for any factor that has inherent uncertainty. Depending on the number of uncertainties (a dynamic input) and the ranges specified for them, our ADET PPM runs a dynamic number of recalculations (ranging from 600 to tens of thousands- based on the input provided) before it is complete. These algorithms enable the computation covering: mean, standard deviation, minimum and maximum in the areas of: requirements points, velocity, defects created, defect points, total sprints and velocity increase. The ADET PPM planning results are recorded in our proprietary APP Tool, included in an overarching Project Workbook, along with actual project data measurements (e.g., actual sprint velocity) taken during project execution. The Project Manager performs statistical analysis of project data against ADET PPB norms, and as needed, uses the ADET PPM to perform what if analysis and re-plan the project.
Collectively these tools, techniques and models bring an industry-first level of planning, insight and execution to agile software development and provides our government sponsors with predictive insights into all software development performance.
For more information regarding these capabilities, please contact us.