Select measures and analytic techniques to be used in quantitative management.
Example Work Products
- Definitions of measures and analytic techniques to be used in quantitative management
- Traceability of measures back to the project’s quality and process performance objectives
- Quality and process performance objectives for selected subprocesses and their attributes
- Process performance baselines and models for use by the project
1. Identify common measures from the organizational process assets that support quantitative management.
Product lines or other stratification criteria can categorize common measures.
2. Identify additional measures that may be needed to cover critical product and process attributes of the selected subprocesses.
In some cases, measures can be research oriented. Such measures should be explicitly identified.
3. Identify the measures to be used in managing subprocesses.
When selecting measures, keep the following considerations in mind:
- Measures that aggregate data from multiple sources (e.g., different processes, input sources, environments) or over time (e.g., at a phase level) can mask underlying problems, making problem identification and resolution difficult.
- For short-term projects, it may be necessary to aggregate data across similar instances of a process to enable analysis of its process performance while continuing to use the unaggregated data in support of individual projects.
- Selection should not be limited to progress or performance measures only. “Analysis measures” (e.g., inspection preparation rates, staff member skill levels, path coverage in testing) may provide better insight into process performance.
4. Specify the operational definitions of measures, their collection points in subprocesses, and how the integrity of measures will be determined.
5. Analyze the relationship of identified measures to the project quality and process performance objectives and derive subprocess quality and process performance objectives that state targets (e.g., thresholds, ranges) to be met for each measured attribute of each selected subprocess.
- Maintain a code review rate between 75 to 100 lines of code per hour
- Keep requirements gathering sessions to under three hours
- Keep test rate over a specified number of test cases per day
- Maintain rework levels below a specified percent
- Maintain productivity in generating use cases per day
- Keep design complexity (fan-out rate) below a specified threshold
6. Identify the statistical and other quantitative techniques to be used in quantitative management.
In quantitative management, the process performance of selected subprocesses is analyzed using statistical and other quantitative techniques that help to characterize subprocess variation, identify when statistically unexpected behavior occurs, recognize when variation is excessive, and investigate why. Examples of statistical techniques that can be used in the analysis of process performance include statistical process control charts, regression analysis, analysis of variance, and time series analysis.
The project can benefit from analyzing the performance of subprocesses not selected for their impact on project performance. Statistical and other quantitative techniques can be identified to address these subprocesses as well.
Statistical and other quantitative techniques sometimes involve the use of graphical displays that help visualize associations among the data and results of analyses. Such graphical displays can help visualize process performance and variation over time (i.e., trends), identify problems or opportunities, and evaluate the effects of particular factors.
- Box and whiskers plots
- Run charts
- Ishikawa diagrams
- Tally sheets
- Classification schemas (e.g., Orthogonal Defect Classification)
7. Determine what process performance baselines and models may be needed to support identified analyses.
In some situations, the set of baselines and models provided as described in Organizational Process Performance may be inadequate to support quantitative project management. This situation can happen when the objectives, processes, stakeholders, skill levels, or environment for the project are different from other projects for which baselines and models were established.
As the project progresses, data from the project can serve as a more representative data set for establishing missing or a project specific set of process performance baselines and models.
Hypothesis testing comparing project data to prior historical data can confirm the need to establish additional baselines and models specific to the project.
8. Instrument the organizational or project support environment to support collection, derivation, and analysis of measures.
This instrumentation is based on the following:
- Description of the organization’s set of standard processes
- Description of the project’s defined process
- Capabilities of the organizational or project support environment
9. Revise measures and statistical analysis techniques as necessary.