Learning from History to Improve Effort Estimation

Poster

Abstract: Creating good effort estimations for software projects is notoriously difficult. An ethnographic field study was conducted at a small web consulting company to better understand the mechanisms by which effort estimates are generated. It was found that the three main sources of error were: 1) imperfect information about the technology and the problem; 2) adding features over the course of the project; and 3) management modification of estimates to win projects. This combination of factors led to two problems: programmers received little feedback regarding the accuracy of their estimates and they lacked the means to track their own estimates that could be used to substantiate the estimates. In addition, projects that had exceptionally accurate estimates were examined for effective practices that could be applied elsewhere. Two such techniques were found: re-estimation and estimation by parts. Lessons learned from this study are currently being applied in the development of a tool to support effort estimation and tracking.

Bio: Wendy Wong is a senior in the School of Information and Computer Science at the University of California, Irvine. Her research, under the guidance of Prof. Susan Elliot Sim, involves designing and implementing an effort estimation and time tracking tool as an Eclipse plug-in.