Software Productivity Measurement and Prediction Methods: what can we tell about them?

Wladmir Araujo Chapetta,Guilherme Horta Travassos

DOI: https://doi.org/10.5753/sbqs.2016.15136

2016-10-24

Abstract:An adequate way of making software organizations to remain competitive is to ensure their innovative capacity and the continuous increasing of their software process productivity with quality. Indeed, the ability on increasing the software productivity relies on among other issues in the organization's measurement and prediction capacity. Productivity refers to the rate at which a company produces goods, and its observation takes into account the number of people and the amount of other necessary resources to produce such goods. However, it is not clear how productivity can be observed when the product is software. Therefore, this work presents the results of an investigation regarding software productivity measurement and prediction methods. A previous systematic literature review was evolved and re-executed, limited to the year 2013. It allowed the identification of 89 new primary studies evidencing that: (1) ratio-based and weighted factors analyses still represent most of the methods applied to measure, describe and interpret software productivity; (2) 24 factors present evidence of influencing productivity, and; (3) SLOC-based measures, despite the criticism and issues associated with these sort of measurements, are the most common measures used in the studies.