An Empirical Study of Analogy-based Software Effort Estimation

  • Authors:

  • Fiona Walkerden;Ross Jeffery

  • Affiliations:

  • Centre for Advanced Empirical Software Research (CAESAR), School of Information Systems, University of New South Wales, Sydney 2052 Australia;Centre for Advanced Empirical Software Research (CAESAR), School of Information Systems, University of New South Wales, Sydney 2052 Australia

  • Venue:
  • Empirical Software Engineering
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

Conventional approaches tosoftware cost estimation have focused on algorithmic cost models,where an estimate of effort is calculated from one or more numericalinputs via a mathematical model. Analogy-based estimation hasrecently emerged as a promising approach, with comparable accuracyto algorithmic methods in some studies, and it is potentiallyeasier to understand and apply. The current study compares severalmethods of analogy-based software effort estimation with eachother and also with a simple linear regression model. The resultsshow that people are better than tools at selecting analoguesfor the data set used in this study. Estimates based on theirselections, with a linear size adjustment to the analogue‘s effortvalue, proved more accurate than estimates based on analoguesselected by tools, and also more accurate than estimates basedon the simple regression model.