Abordagem de pontos de função no desenvolvimento de sistemas utilizando práticas ágeis

Abstract

Function point analysis is a technique that enables the measurement of software development projects and, thus, establish a size unit that is used in software project management. Currently, agile methods and practices have been widely used and have been gaining space in software development projects. Given that function point analysis is already consolidated and projects using agile methods are not estimated by function points, this work explores the use of function points with agile practices and proposes a hybrid approach called HiPA (Portuguese abbreviation for Hybrid with Agile Practices). In order to present the current state of the art regarding the user of function points in projects adopting agile practices, a systematic mapping of 182 articles was done and 18 of them were selected for this work. With access to information on software development projects from a single organization, this work obtained indicators that allowed an evaluation of such projects, part of them using the traditional waterfall methodology and others adopting agile practices. Those indicators were analyzed, in a case study, to achieve the goal of verifying the feasibility of using function points in projects that use agile practices. The projects analyzed in the case study totaled more than 24,000 hours and the results obtained compare the two above-mentioned software development models (waterfall and agile). Numeric results from the execution of projects in terms of time and cost are presented, being the IDC (Portuguese abbreviation for Cost Performance Index) better for projects that adopt agile practices. Lastly, although the literature does not inform how agile practices should be used with function points, this work proposes a hybrid approach, called HiPA, to software development and evaluates it through a questionnaire, in which 61.5% of the respondents indicated HiPA as a viable approach while 38.5% as possibly viable. The implementation of the HiPA approach with the support of counting function points was partially evaluated, with results pointing the approach as viable.