Identifying Common Groups of Highway Projects Using Fuzzy Cluster Analysis: Fid-Bau Portal

Identifying Common Groups of Highway Projects Using Fuzzy Cluster Analysis

Nguyen, Phuong / Tran, Dan / Lines, Brian

The construction industry has a great source of historical data with various attributes, including both quantitative and qualitative. The historical data can be used to identify common groups (i.e., clusters) of projects that share a high degree of similarity in particular project attributes. The identifying results can create opportunities for a variety of comparisons between clusters that help select the optimal alternatives in different decision-making scenarios. Qualitative variables play an important role in many construction-related decisions but are difficult to mathematically evaluate. The objective of this study was to demonstrate the application of fuzzy sets in the context of fuzzy cluster analysis to identify common groups of projects in a collected highway dataset based on similarities in project attributes. Fuzzy cluster analysis can provide an effective tool to evaluate qualitative data (e.g., rating of experts) that is commonplace in assessment of construction decision-making criteria. This application was conducted in an R environment to provide a process reproducible with any datasets in other decision-making scenarios. The result shows that the number of seven clusters of highway projects makes the most logical sense in the given dataset by using four cluster validity indices. This study contributes to the body of knowledge by demonstrating a reproducible process of applying fuzzy cluster analysis to explore common groups of highway projects that share similarities in facility type, project type, complexity, project risk, delivery methods, and cost growth. Multiple comparisons between clusters’ attributes can be extracted from the clustering results to aid decision-making in construction.