Managing of Subcontractors Failure in Construction Projects.

In a construction project, especially in risky environments, subcontractors are considered as important as the main contractor since most work is carried out by them. A crucial step in building projects is choosing the finest subcontractor. When choosing subcontractors, a variety of considerations must be taken into account. A poor choice of subcontractors might cause a lot of issues when the job is being done. These include subpar work and project length delays. Many things influence this process. From earlier studies, variables that affect the choice of a subcontractor have been compiled. The main aim of the paper is to predict the failure of the subcontractor by using a mix of several advanced technology as BIM, machine leaning and building a system. The methodology of the paper is first to use BIM to see the mistakes and unqualified quantity estimation by the subcontractors which lead to increase the cost of the projects and hence failures of the projects. The second step is to identify primary reasons for subcontractors' failure and identify and evaluate the major reasons why contractors fail on industrial projects in Iraq. The third step is to predict the performance of the subcontractor using three algorithms (logistic regression, decision tree and Naive Bayes) with data collected from 234 projects and see how the wrong selection of the subcontractors leads to incomplete projects in the future. The variables that used in the paper include the most effective one, as change order effect directly on the projects performance as well as cost, time, deviation, type of projects and the rate of project completion, the low the percentage of completion the worse the performance. The final step is to develop a system to manage the subcontract in different projects stage. The results show that there is a different in the quantity estimated between the actual and the estimated range between (5% to 15%) in several item of the building, which consider high. the best algorithms for the is Naive Bayes as give accuracy about 97%. [ABSTRACT FROM AUTHOR]