Lawler and Kitchenham present an automated measurement model that provides a standard method for defining measures and metrics.
Currently, most companies’ measurement programs suffer both from a lack of metric standards, and from invalid and missing data. This reduces data comparability and confidence in report validity. The authors propose a measurement model that is automated by the TychoMetrics commercial tool. The proposed model and tool address the following components of a measurement program: definition of measurements to collect and metrics to calculate; links between measures and development tools; automatic interrogation of development tools; and data extraction, instantiation, storage, collation, and reporting.
This paper describes how to define a measurement protocol that allows full automation of the measurement process. The authors present the measurement protocol’s static elements: atomic measures; measurement sources that include one or more atomic measures; measurement schemas that model relationships between measurement groups within a particular measurement source; and measurement object models that result from combining two or more measurement schemas, or portions of schemas. The TychoMetrics tool instantiates the measurements’ object models with actual data values, from the company automating the calculation of metrics and generation of reports.
Recent software engineering trends have been pushing companies to adopt formal measurement programs. Despite this need, existing programs do not deliver currently required capabilities. This work provides a consistent measurement model that ensures consistent data collection, which in turn increases the validity of the data analysis results. The work delivers on its promise, since it finds a more rigorous approach to data collection than the ad hoc system used today by most organizations. This is an example of a reliable and rigorous work.