A novel method for forecasting Construction Cost Index based on complex network

摘要

Construction Cost Index (CCI) forecasting has been paid great attention by civil engineers and cost analysts for its broad applications in construction industry. In this paper, for more accurate predictions of CCI, a novel method is proposed based on the analysis of complex network. Firstly, CCI data are mapped into a network by visibility graph. Afterwards, the link prediction method is combined to determine the node similarity in the network. Then initial predictions are made based on the analysis of node similarity. Finally the node distance is taken into account to improve the preliminary forecasting results. In the CCI prediction experiment, we illustrate the applicability and predictability of our method by error comparison and t test. It is believed that the method is able to predict CCI more accurately, which can contribute to saving costs and making budgets in construction industry. (C) 2019 Elsevier B.V. All rights reserved.