IEEE TRANSACTIONS ON SOFTWARE ENGINEERING,
Software Productivity Measurement Using Multiple Size Measures
Barbara Kitchenham and Emilia Mendes, Member, IEEE Computer Society
Abstract—Productivity measures based on a simple ratio of product size to project effort assume that size can be determined as a single measure. If there are many possiblesize measures in a data set and no obvious model for aggregating the measures into a single measure, we propose using the expression AdjustedSize/Effort to measure productivity. AdjustedSize is defined as the most appropriate regression-based effort estimation model, where all the size measures selected for inclusion in the estimation model have a regression parameter significantly different fromzero (p < 0.05). This productivity measurement method ensures that each project has an expected productivity value of one. Values between zero and one indicate lower than expected productivity, values greater than one indicate higher than expected productivity. We discuss the assumptions underlying this productivity measurement method and present an example of its use for Web application projects.We also explain the relationship between effort prediction models and productivity models. Index Terms—Software productivity measurement, software cost estimation.
the amount of output (what is produced) per unit of input used. In general, productivity is difficult to measure because outputs and inputs are typically quite diverse and are oftenthemselves difficult to measure. In the context of software, productivity measurement is usually based on a simple ratio of product size to project effort (e.g., ). Thus, if we can measure the size of the software product and the effort required to develop the product, we have: P roductivity ¼ Size=Effort: ð1Þ
Equation (1) assumes that size is the output of the software production process andeffort is the input to the process. This can be contrasted with the viewpoint of software cost models where we use size as an independent variable (i.e., an input) to predict effort which is treated as an output. Equation (1) is simple to operationalize if we have a single dominant size measure, for example, product size measured in lines of code. MacCormak et al. used new lines of code per person dayin a recent productivity study, noting that, “It is an imperfect measure of productivity but one that could be measured consistently” . However, there are circumstances when there are several different effort-related size measures and there is no standard model for aggregating these measures. We
. B. Kitchenham is with the National ICT Australia, Locked Bag 9013, Alexandria, NSW 1435,Australia, and with the Department of Computer Science, Keele University, Staffordshire, ST5 1BG, UK. E-mail: Barbara.Kitchenham@nicta.com.au, Barbara@cs.keele.ac.uk. . E. Mendes is with the Computer Science Department, University of Auckland, Private Bag 92019, Auckland, New Zealand. E-mail: firstname.lastname@example.org. Manuscript received 9 Mar. 2004; revised 17 Aug. 2004; accepted 10 Nov. 2004.Recommended for acceptance by A. Mili. For information on obtaining reprints of this article, please send e-mail to: email@example.com, and reference IEEECS Log Number TSE-0037-0304.
0098-5589/04/$20.00 ß 2004 IEEE
recently collected data from Web companies where effort was strongly correlated with several different size measures (e.g., number of new Web pages, number of high effort functions, number ofnew images), each of which measured a different aspect of the overall size of a Web application and contributed significantly to a regression-based effort prediction model. When we have a number of significant size measures related to effort, it is difficult to determine how to construct a single size measure from the different individual measures. This means we cannot use (1) for productivity...
Leer documento completo
Regístrate para leer el documento completo.