Data Preprocessing Method for Cost Estimation of Building Projects

Abstract

All construction projects have unique characteristics that must be considered during cost estimating and checking activities. Especially in the conceptual and schematic design stages, owners require precise cost estimates than their information providing. However, information related to the project scope is more likely to change in the early design phases in response to ongoing scope changes. Also, as only minimal project scope information is available during the early stages of estimation, cost estimators require effective estimation strategies. In practice, parametric cost estimation—which utilizes historical cost data—is the most commonly used method in these initial phases. Therefore, compilation of historical data pertaining to appropriate cost variance governing parameters is a prime requirement. However, data mining (data preprocessing) for denoising internal errors or abnormal values must be performed before this compilation. To address these issues, this research utilizes applied statistical methods, and it develops an alternative cost model based on these. To construct the model, the building cost data of 124 apartment projects, which are selected for case study, in Korea are compiled.. The main objectives of the suggested model are to effectively prepare strategic and conceptual cost estimates, and to provide check and control functions during the conceptual and schematic design stages. The cost model is expected to provide more accurate and stable estimates than the conventional methods.