ENNA: Software effort estimation using ensemble of neural networks with associative memory

Yigit Kultur, Burak Turhan, Ayse Basar Bener

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

19 Citations (Scopus)

Abstract

Companies usually have limited amount of data for effort estimation. Machine learning methods have been preferred over parametric models due to their flexibility to calibrate the model for the available data. On the other hand, as machine learning methods become more complex they need more data to learn from. Therefore the challenge is to increase the performance of the algorithm when there is limited data. In this research we used a relatively complex machine learning algorithm, neural networks, and showed that stable and accurate estimations are achievable with an ensemble using associative memory. Our experimental results revealed that our proposed algorithm (ENNA) achieves on the average PRED(25) = 36.4 which is a significant increase compared to Neural Network (NN) PRED(25) = 8.

Original languageEnglish
Title of host publicationSIGSOFT 2008/FSE-16 - Proceedings of the 16th ACM SIGSOFT International Symposium on the Foundations of Software Engineering
Pages330-338
Number of pages9
DOIs
Publication statusPublished - 1 Dec 2008
Externally publishedYes
Event16th ACM SIGSOFT International Symposium on the Foundations of Software Engineering, SIGSOFT 2008/FSE-16 - Atlanta, GA, United States of America
Duration: 9 Nov 200814 Nov 2008

Conference

Conference16th ACM SIGSOFT International Symposium on the Foundations of Software Engineering, SIGSOFT 2008/FSE-16
Country/TerritoryUnited States of America
CityAtlanta, GA
Period9/11/0814/11/08

Keywords

  • Adaptive Resonance Theory
  • Associative Memory
  • Bootstrap
  • Cost Estimation
  • Effort Estimation
  • Ensemble
  • K nearest neighbors
  • Multilayer Perceptron
  • Neural Network

Cite this