Research On Construction Agency Selection In Public Works Based On The Performance Prediction

Selecting the ideal construction agency is one of the most important decisions of the ultimate owners (the government) in the agent construction projects. Meanwhile, this topic has attracted a lot of attentions from the researches in both theory and engineering fields. In view of the agent selection problem in the existing research and practice, the paper focuses on the agent selection under the complex project situation which has the characteristics of the whole process of enterprise-type agent construction. The ideas and methods of performance prediction will be creatively applied in the overall improvement of agent selection’s methods, indicators and paths.Based on the proposed basic analytical framework of performance predictions, the main research contents and results are as follows:(1) The paper identifies the agent project performance’s drivers and components through the theoretical analysis and analyzes the relationship between the both by SEM; the agent capacity dimension, the external environment dimension and the system of the agent project performance’s components was established as the basis of performance prediction. Moreover, a performance evaluation method based on fuzzy mathematics and its implementation steps were given.(2) Based on the prediction theory, paper builds the conceptual model can reflect the nonlinear mapping relationship between six functional group variables (including project management skills, technical capacity, economic financial capacity, resource capacity, the external environment and project properties) and the project performance of agent. This paper also presents a predictive index system of the agent project performance and designs a set of complete predictive model of BP neural network. Finally, the agent project performance prediction based on MATLAB toolbox is given by programming.(3) Based on the practice of international engineering consulting service procurement, the article puts forward a new path of agent selection based on the performance prediction: qualification examinationâ†'agent list of dynamic managementâ†'preparation before bid opening and bid openingâ†'postqualificationâ†'performance predictionâ†'interviewâ†' interview scores plus performance prediction scoresâ†'negotiationâ†'awarded construction contract. Finally, the paper puts forward ten indicators and a discriminant method to distinguish simple and complex projects. When the proposed project is a complex project, selection methods and paths based on the performance prediction would help owner to choose the most suitable agent.