Drivers for cost estimating in early design: Case study of residential construction

This paper proposes the use of a series of independent variables for an early estimation of the building construction cost of residential buildings. Based on 70 German residential properties, these variables serve as the cost drivers of a project, and the regression model, tested against the 70 properties, has a mean absolute percentage error of 9.6%. When applied to predict the cost of five more properties that were excluded from the 70 in the regression model, the percentage error ranges between-12 and 13%. The identified cost drivers are: compactness of the building, number of elevators, size of the project, expected duration of construction, proportion of openings in external walls, and region.