Abstracts – Browse Results

Search or browse again.

Click on the titles below to expand the information about each abstract.
Viewing 1 results ...

Chen, X, Zhang, Y, Zhao, B and Yang, S (2021) Investment Probabilistic Interval Estimation for Construction Project Using the Hybrid Model of SVR and GWO. Journal of Construction Engineering and Management, 147(05).

  • Type: Journal Article
  • Keywords: Cost prediction; Support vector regression (SVR); Confidence interval; Nonparametric estimation; Grey wolf optimization (GWO); Error distribution;
  • ISBN/ISSN: 0733-9364
  • URL: https://doi.org/10.1061/(ASCE)CO.1943-7862.0002032
  • Abstract:
    Investment estimation is a key component of early decision-making for a construction project, which is crucial to the project cost control. Currently, most investment estimation researches render the point value results, which could lead to considerable uncertainty in the estimation results and increase the risk of decision-making. Therefore, it is essential to explore a type of systematic, accurate, and effective estimation method. This study proposed an innovative estimation method of probability interval prediction based on the distribution of prediction errors. First, the dimension reduction of the construction indexes was conducted by using exploratory factor analysis (EFA). Then, a model was developed based on the fusion of the support vector regression (SVR) and grey wolf optimization (GWO) algorithm. Finally, cost intervals with different confidence levels were obtained on the basis of kernel density estimation (KDE). The case results indicated that when the confidence was 95%, the comprehensive evaluation index coverage width-based criterion (CWC) and the interval coverage rate PICC of the cost estimation were 2.17 and 93.33%, respectively. Hence, the proposed interval prediction model was fairly reliable, which could provide practical guidance for the investment decisions in the early stage of construction projects and give the decision makers more abundant forecasting information.