Preliminary communication

Model for cost prediction of prefabricated housing

Mladen Vukomanović orcid.org/0000-0002-3037-1908 ; University of Zagreb, Faculty of Civil Engineering, Kačićeva 26, 10000 Zagreb, Croatia
Mirsad Kararić ; Libra projekt d.o.o., Siget 18c, 10000 Zagreb, Croatia

Abstract

The paper analyzes the application of neural networks on cost prediction of prefabricated housing, on more than 30 buildings. 17 variables have been identified in the prediction model, with the following performances: 83,8%of predicted values had the deviation lower than 5%and 12,9%had the deviation lower than 10 %, in relation to actual values. The model was verified on 28 and validated on 3 buidings, with the deviation of 4,6%against actual values.With thismodel contruction companies can, during tendering phase, effectively predict total cost of prefabricated housing and thus acquire competitive advantage in the market. Conclusion brings guidelines for the model's use in practice and gives recommendation for its further development.

Keywords

building cost; prefabricated housing; prediction; model; neural networks

Hrčak ID:

40963

URI

https://hrcak.srce.hr/40963

Publication date:

30.9.2009.

Article data in other languages: croatian

Visits: 2.536 *