Task Effort Fuzzy Estimator for Software Development

Abstract

Software estimation has been one of the biggest challenges in Computer Science for the last decades. This practice is essential for research and development companies, since it can provide cost and deadline forecasting, for example. Most of the traditional techniques such as function points, regression models, COCOMO, etc, require a long-term estimation process, which is unworkable for greatly dynamic companies who demand short-term task estimation, that is, in days. Furthermore, such techniques need historical data from past similar tasks, which may not exist when estimating an original task, a common situation for companies developing high technology. A well-known technique for short-term task effort estimation is the Use Case Points method, which only needs Use Case documents to start the estimation process. Based on this context, this work aims to apply the Fuzzy systems theory to estimate the effort for original software tasks in short-term, in a research and development company, based on the Use Case Points technique. Information from a real software project is provided for this methodology and used to validate the results. Results are validated by comparing the tasks planned with the Fuzzy estimator and the Use Case Points technique, and the expended execution time.