Estimating Software Project Effort by Analogy Based on Linguistic Values

  • Authors:

  • Ali Idri;Alain Abran;Taghi M. Khoshgoftaar

  • Affiliations:

  • -;-;-

  • Venue:
  • METRICS '02 Proceedings of the 8th International Symposium on Software Metrics
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

Estimation models in software engineering are used topredict some important attributes of future entities such assoftware development effort, software reliability andprogrammers productivity. Among these models, thoseestimating software effort have motivated considerableresearch in recent years. The prediction procedure usedby these software-effort models can be based on amathematical function or other techniques such asanalogy based reasoning, neural networks, regressiontrees, and rule induction models. Estimation by analogy isone of the most attractive techniques in the software effortestimation field. However, the procedure used inestimation by analogy is not yet able to handle correctlylinguistic values (categorical data) such as 'very low','low' and 'high'. In this paper, we propose a new approachbased on reasoning by analogy, fuzzy logic and linguisticquantifiers to estimate software project effort when it isdescribed either by numerical or linguistic values; thisapproach is referred to as Fuzzy Analogy. This paper alsopresents an empirical validation of our approach based onthe COCOMO'81 dataset.