Classification of building project complexity and evaluation of supervisory staffing patterns using cluster and factor analysis techniques

Abstract

The impacts of complexity of building projects on project performance, project success and project staffing, etc. have been well acquiesced by the construction industry. This research describes the classification and evaluation of building projects and supervisory staffing patterns by using natural clustering and factor analysis techniques. The prime objectives of this study include the re-grouping of the variables affecting the complexity of construction by factor analysis and the classification of building projects into distinguished nine project clusters by Hierarchical Cluster Analysis (HCA). Subsequently, the supervisory staffing patterns of the building project samples have been examined and analysed. Previous researchers usually adopted subjective measures in studying various behaviour patterns on building projects. Sixty variables affecting the complexity of construction have been identified and 62 project samples have been collected for the study. Twenty-seven variables have been selected for building up a 6-factor model by using factor analysis. The project factor scores were used for classifying the building project samples into a 9-cluster project model by HCA. The resultant project clusters were well structured and the characteristics of building projects could be defined. The project clusters have been examined and affirmed by using K-Mean Cluster Analysis, Discriminant Analysis and Kohonen Self-organization Map Algorithm neural network. Subsequently, a Construction Complexity Index?(CCI) has been developed to provide an objective tool in measuring the complexity of construction for building projects in Hong Kong. The staff costs of the project clusters have been studied and the differences in staffing patterns are addressed in this report. It is found that staff costs might not be allocated in line with the complexity of construction. There were significant differences between staffing strategies for different types of building projects among contractors. The findings explained possible reasons for differences in project performance and project success. Subsequently, the validity of factor model, project cluster model and CCI were affirmed by structured interviews with experts in construction management from contractors. When comparing the supervisory staffing patterns, significant differences were found between recommendations of the experts and the contractors in pricing of supervisory staff especially the staff costs for middle management. In summary, this study provides objective views on the complexity of construction for building projects of which the valuable information would enable senior management to assign supervisory staff concerned. The project clusters and project groups identified and the CCI derived are unbiased and can be used as yardsticks for the industry in Hong Kong or similar grouping can be built up internally by contractors for facilitating pricing and staffing works.

Date of Award	2 Oct 2007
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Chi Ming TAM (Supervisor)