Quantification of Cost Uncertainties Using Neural Network Technique

See related content

by Awad S. Hanna, Univ of Wisconsin-Madison, Madison, United States,
Li-Chung Chao, Univ of Wisconsin-Madison, Madison, United States,

Document Type: Proceeding Paper

Part of: Computing in Civil Engineering:

Abstract: This paper describes the development of a neural network system for forecasting cost escalation in construction. The input of the neural network system consists of factors that have an impact on cost escalation, such as inflation rates for labor and material, and rate of change in construction demand. The output is the annual rate of construction cost escalation. Historical data for the past 25 years (1968-1992) obtained from Engineering News Record is organized into example input-output patterns. The prepared patterns for the first 15 years are used for training the neural network. The rest of them are withheld for testing the performance of the trained network in terms of prediction errors. Test results show that the forecasts produced are satisfactory considering the complexity and irregularity involved. Discussions and suggestions for future works are also provided.

Subject Headings: Construction costs | Neural networks | Forecasting | Errors (statistics) | Computer models | Construction materials | Uncertainty principles |

Services: Buy this book/Buy this article


Return to search