Model Experimentation for Noncompleted Construction Project Cost Forecasting Methods

摘要

It is significant for us to achieve the cost control targets of non-completed project by using project cost forecasting models. After reviewed some literatures, the author of this paper firstly analyzed the theories, methods and forecasting process of the unitary linear trend forcasting model, the logistic curve model and the GM(1, 1) model. And then, a experimentation has been carried out to test the fitting precision and forecasting precision of the 3 forecasting models. Experimentation results shows that: the Logistic curve model can perfectly reflect the law of the project cost, and can obtain a high fitting and forecasting precisions except the high fitting deviations in initial stages of the construction processes. It is favorable for us to use the GM(1, 1) model in the initial stages of the project when data is not adequate becease it can get a high forecasting precision by few data of the system without testing its statistic characteristic quantities. It is not judicious for us to use the unitary linear trend forcasting model to forecast the non-completed project cost because its fundamental hypothesis is against the law of the project cost.