학술논문
Cost modelling of office buildings in Hong Kong: an exploratory study
Document Type
Academic Journal
Author
Source
Facilities, 2005, Vol. 23, Issue 9/10, pp. 438-452.
Subject
Language
English
ISSN
0263-2772
Abstract
This paper presents a set of step‐wise regression models which can incorporate multiple factors in modelling the costs of office buildings. The models appeared to be more accurate than the traditional method.
The data were collected from historical office building projects, which were then, adjusted using the construction price index. The step‐wise regression was conducted to produce the linear cost models.
Seven RC office buildings and 11 steel office buildings in Hong Kong completed in different years were selected randomly to verify the accuracy of the regression models developed. The data of these buildings were not used in the development of the cost models. The result shows that the variability of percentage difference is ranging from −4.11 per cent (4.11 per cent underestimate) to +2.74 per cent (2.74 per cent overestimate) for RC office buildings. For steel office buildings, it ranges from −6.65 per cent (6.65 per cent underestimate) to +2.78 per cent (2.78 per cent overestimate).
This study presents a methodology that can be used in cost estimation of office buildings in Hong Kong at early stage of construction project. The regression cost models developed above are based on, in total, historical data of 30 completed office buildings in Hong Kong. The reliability of the cost models can be further improved by including more office buildings to develop the cost models. Furthermore, the application of cost modelling by regression analysis is not limited to office buildings. The same approach can be applied to residential and other non‐residential buildings as well. Regression cost modelling, with sufficient updating for new cost data available, can provide economic, quick and accurate cost estimation at early stage of construction projects. It will become rational guide supplementing judgmental forecast of cost advisors in near future.
Step‐wise regression procedure was applied to develop the cost models. Jackknife re‐sampling was carried out and both of the models show stability. Cross‐validation shows that the developed regression models performed satisfactorily. The paper considers that it can provide economic, quick and accurate cost estimation at the early stage of construction project. In addition, the approach of this study can be adopted to develop cost models of other types of buildings in other locations.
The data were collected from historical office building projects, which were then, adjusted using the construction price index. The step‐wise regression was conducted to produce the linear cost models.
Seven RC office buildings and 11 steel office buildings in Hong Kong completed in different years were selected randomly to verify the accuracy of the regression models developed. The data of these buildings were not used in the development of the cost models. The result shows that the variability of percentage difference is ranging from −4.11 per cent (4.11 per cent underestimate) to +2.74 per cent (2.74 per cent overestimate) for RC office buildings. For steel office buildings, it ranges from −6.65 per cent (6.65 per cent underestimate) to +2.78 per cent (2.78 per cent overestimate).
This study presents a methodology that can be used in cost estimation of office buildings in Hong Kong at early stage of construction project. The regression cost models developed above are based on, in total, historical data of 30 completed office buildings in Hong Kong. The reliability of the cost models can be further improved by including more office buildings to develop the cost models. Furthermore, the application of cost modelling by regression analysis is not limited to office buildings. The same approach can be applied to residential and other non‐residential buildings as well. Regression cost modelling, with sufficient updating for new cost data available, can provide economic, quick and accurate cost estimation at early stage of construction projects. It will become rational guide supplementing judgmental forecast of cost advisors in near future.
Step‐wise regression procedure was applied to develop the cost models. Jackknife re‐sampling was carried out and both of the models show stability. Cross‐validation shows that the developed regression models performed satisfactorily. The paper considers that it can provide economic, quick and accurate cost estimation at the early stage of construction project. In addition, the approach of this study can be adopted to develop cost models of other types of buildings in other locations.