Estimating design effort using a neural network methodology

Hamdi A. Bashir*, Ahmed El-bouri, Vince Thomson

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

Design effort estimation is an essential process to estimate the final cost as well as the duration of a future design project. This paper describes the application of an artificial neural network to design effort estimation. Using historical data from General Electric (GE) Hydro†, a General Regression Neural Network (GRNN) model is constructed to learn the relationship between four major factors, namely, product complexity, technical difficulty to team expertise ratio, type of drawings submitted to the customer, and involvement of design partners as inputs, and effort needed to execute designs for hydroelectric generators as an output. The performance of the model was evaluated in terms of a number of well known objective criteria. The results indicate that artificial neural networks can be considered as a promising tool for improving the estimation accuracy of design effort, and consequently minimizing the probability of schedule and cost overruns. Significance: Delivery of products as scheduled and within budget is one aspect of measuring the success of a project. This paper presents a methodology that provides greatly improved design effort estimates. This is of interest to industry since improving design effort estimation accuracy is a major factor in minimizing the severity of cost overrun and schedule slippage.

Original languageEnglish
Pages (from-to)341-348
Number of pages8
JournalInternational Journal of Industrial Engineering : Theory Applications and Practice
Volume13
Issue number4
Publication statusPublished - Dec 2006
Externally publishedYes

Keywords

  • Artificial neural networks
  • Design project
  • Effort estimation

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering

Fingerprint

Dive into the research topics of 'Estimating design effort using a neural network methodology'. Together they form a unique fingerprint.

Cite this