Applying Estimation and Metrics in CAMS


CAMS encourages the creation and application of metrics in the form of a quality strategy of the organization. Quality management, assurance, and testing process maps provide the basis for an overall quality approach in an organization. Identification and application of estimation and metrics need to be coordinated across multiple teams, projects, and time (i.e., previous project metrics data need to be used in estimating current and future projects). Galorath (2009) has described major steps and related activities in implementing metrics. The following are the steps that can be used in applying metrics in CAMS-based projects:

- Identify the purpose of estimation: Define and document the purpose of estimations and the expectation from the data. A baseline can be created to include planning goals, initial business requirements, and so on, to ascertain what the metrics will achieve. Once the purpose becomes clear, relevant metrics can be created and used.

- Agree on the rules for metrics: Recognize constraints associated with the project and related applications that affect metrics and measurements.

- Decide who will measure and what will be measured: Responsibilities for metrics must be assigned to the roles. For example, the developer can measure the person-days spent on a feature, whereas the project manager focuses on the project velocity. This should include trained and skilled people, proper technology, and tools as well as a documented and repeatable estimation process.

- Evaluate metrics: Subject the metrics themselves to a quality evaluation. Are the metrics still fit for the purpose for which they were created? Organizations should use a metric evaluation checklist in order to confirm the use and purpose of every metric (Hartmann and Dymond, 2006). Ensure that the data associated with a metric is properly collected, applied methods are efficient and verified, identified results are precise, and the quality is properly maintained. Metrics evaluation should make sure that adopted ground rules are consistently applied throughout the estimation process and all the related assumptions are properly assessed beforehand.

- Separate metrics from measurements (data): Ascertain the metrics in a collaborative manner, but then separate the metrics from the process of collecting the data. Data collection can become an important exercise that requires use of automated tools or manual entry of information within a project. A schedule for data collection needs to be determined and defined.

- Create a baseline: This is the starting point of the use of metric. A baseline bounds the metric collection program and thereby facilitates analysis and trend spotting. This provides the time boundary for data collection. This also enables analysis of data based on the timeline in subsequent projects.

- Determine metrics risks: Risks in metrics are not only the project risks but also the risks associated with understanding and applying metrics. For example, an incorrect suite of data may get inadvertently analyzed and used in estimating new tasks for a team. This risk has to be factored in and avoided through accuracy of measurements. The impact of a risk can be reduced and even avoided through accuracy in data collection.

- Estimate validation and review process: The techniques of V&V can also be applied to the estimates made in a project. This is a collaborative effort to cross-check the accuracy of data used in estimation, potential risks associated with dynamically changing situations, and subjective factors that may not have been easily quantified within the metrics.

- Post project reviews of metrics and documenting lessons learned: Each time a particular metric is used in practice, identify its accuracy and impact. This is usually done at the end of a project during the post-project review. This documentation on how a metric was used provides valuable information on its efficacy. Lessons learnt in the process can be used in further estimates in a project. This documentation also provides validity of the process used in estimation. Actual results versus earlier estimations can be standardized for future estimates.

Taken from : The Art of Agile Practice: A Composite Approach for Projects and Organizations


Subscribe to Computing Tech

Enter your email address:

Delivered by FeedBurner

Add to Technorati Favorites Top Blogs