Implementation of Fuzzy Cluster Analysis to Partition Cost Performance into Typical Groups Based Upon Project Characteristics Phuong H.D. Nguyen, Ph.D. Student, Dan Tran, Ph.D., and Brian C. Lines, Ph.D. University of Kansas Lawrence, Kansas Quantification of project cost performance plays an essential role in any decision-making process in any transportation infrastructure projects. Project characteristics, such as facility type, project type, and complexity are critical factors that certainly affect predictions and classifications of cost performance. Due to the fact that project complexity and other project attributes are often given in terms of subjective judgements and ratings of experts, it is difficult to quantify those qualitative data types. This research implemented the fuzzy set theory in the context of fuzzy cluster analysis to classify/partition input data of cost performance and project characteristics into meaningful groups. The fuzzy classification process was conducted with a dataset of 254 horizontal transportation projects collected by the Federal Highway Administration (FHWA) in 2012. As a result, this paper shows the applicability of fuzzy cluster analysis within the construction industry with common cluster validity indices. This research contributes to the construction body of knowledge and practitioners a new method to classify cost performance data and understand its underlying structures and behaviors. Future work of this study is to examine other project performance measurement metrics, such as schedule performance and quality. Key words: fuzzy set theory, fuzzy c-means cluster analysis, cost performance, project complexity Introduction Cost performance is one of the key criteria for construction project owners and stakeholders to consider in any decision-making process. Determination of cost performance is varied on a project-to-project basis and depends on many factors, including project characteristics, internal and external conditions, and other features (Baccarini 2004). During feasibility, planning, and design stages, cost estimates are prepared in detail based upon project characteristics and take into account potential costs for associated project uncertainty and complexity (Creedy et al. 2010). Due to the fact that different levels of project complexity drive the estimated cost performance diversely, it is important that both industry practitioners and academic researchers need to generate empirical and sufficient methods/frameworks/models to quantify cost performance based on project characteristics. Traditionally in construction, cost performance is measured and estimated based on probabilistically mathematical frameworks and simulation models (Creedy 2006, Dikmen et al. 2007). Among the factors that affect cost performance, levels of project complexity and uncertainty play an important role in establishing more deterministically predicting results (Hegazy and Ayed 1998). However, uncertainty and complexity are often given in terms of linguistic input variables. In other words, those factors are assessed based on experience of professionals and experts in the context of subjective judgements. One of the weaknesses of probability theory and simulation methods is not able to quantify qualitative input data (Hastie et al. 2009). Despite the fact that those methods are often considered to quantity project characteristics and other attributes, a more comprehensive method, which can incorporate quantifications of both quantitative and qualitative input data, is necessary. Hegazy and Ayed (1998) declared that fuzzy set theory is useful in the construction industry where realistic historical project data are limited. In the field of engineering, fuzzy set theory has been used to capture qualitative domain professional judgements to generate theoretical decision-making models and widely applied to many areas, such as computer science, mechanical engineering, aerospace engineering, and chemical engineering (Elwood 2014, Kruse et al. 2007). As a promising method to mathematically take into account the subjective judgements and expressions of construction professionals and experts about project complexity, fuzzy set theory was applied in this study in terms of a soft clustering method introduced by Bezdek (1993). This method is also known as the fuzzy c-means (FCM).