Font Size: a A A

Study On Selecting The Project Delivery System Based On Random Forest

Posted on:2022-02-09Degree:MasterType:Thesis
Country:ChinaCandidate:M W LiFull Text:PDF
GTID:2492306569965439Subject:Architecture and Civil Engineering
Abstract/Summary:PDF Full Text Request
Selecting an appropriate project delivery system is an important decision made from the owner in the pre-project stage,and it is considered as one of the crucial factors for project success,so it is of great practical significance to research on the selection of project delivery system.This thesis tries to establish the influencing factor system of the selection of project delivery system and to build the decision support system for project delivery system based on the random forest algorithm by utilizing project data.It aims to extend the existing research ideas and methods of project delivery system selection from theoretical point of view,and help owners to select suitable project delivery system in practice.The main research of this paper includes three parts.Firstly,constructing suitable influencing factor system of project delivery system.This paper reviews the research on the selection of project delivery system in the past and identifies the influencing factors.Subsequently,Mean-score and Ranking Agreement Factor methods are used to screen the identified influencing factors and construct the index system of influencing factors of project delivery system selection.Finally,the influence factor index system is analyzed and the calculation method of each index is determined.Secondly,the decision support system is constructed.In this paper,for the selection between traditional transaction mode and engineering general contracting transaction mode,51 engineering project data are collected and random forest algorithm is introduced to construct the decision model of project delivery system.After testing,the out-of-bag data error value OOB_Error of the model is 0.211,the prediction accuracy is 71.3%,the geometric mean G-Mean is 0.707,and the Kappa coefficient is 0.441,which indicates that the model has a more desirable practical reference value.Subsequently,this paper ranked the importance of influencing factor indicators based on the mean impurity decline(MDI),and found that in the selection of these two types of transaction methods,the normalized weights of five factors,namely,owners expected participation,project design complexity,project requirements for cost certainty,owners management capability a nd project requirements for schedule certainty,were 11.47,11.22,10.33,9.50 and 9.21,which have significant influence on the selection of project delivery system and are the indicators that owners need to give priority to when making decisions on the selection of project delivery system.Subsequently,the case study is conducted on the decision support system.This paper takes a school-constructing-project as an example and applies the decision support system proposed in this paper to make the decision of project delivery system selection to verify the operability of the model.It is found that the decision results obtained from the model match with the engineering reality,which indicates that the model has certain feasibility.Subsequently,by analyzing the problems that appear in the decision-making process,the shortcomings of the model are pointed out and improvement suggestions are put forward.The selection of project delivery system is an overly complex and interesting research topic,and this paper is only a preliminary exploration,and it is believed that more and more valuable research results will emerge in the future.
Keywords/Search Tags:Project Delivery System, Random Forest, Decision Support System, Influencing Factors Importance Ranking
PDF Full Text Request
Related items