Issue No. 03 - May/June (2003 vol. 18)
Gary D. Boetticher , University of Houston-Clear Lake
<p>Corporations invest over 300 billion dollars annually in software production. A key question in the software development process is, When will it be done? Estimating techniques include human-based (expert and analogy), algorithmic (Function Point Analysis and Cocomo [Cost Constructive Model], and machine learner-based. Human-based techniques are the most popular. However, machine learner-based techniques have generated impressive results, including accuracy rates within 25 percent, 83 percent of the time in the software life cycle's early stages. This article presents details from three machine learner successes in software effort estimation.</p>
machine learners, software engineering, project estimation, neural networks, backpropagation, software metrics, Bayesian analysis, effort estimation, programming effort
G. D. Boetticher, "When Will It Be Done? Machine Learner Answers to the 300-Billion-Dollar Question," in IEEE Intelligent Systems, vol. 18, no. , pp. 48-50, 2003.