| description abstract | State highway agencies (SHAs) frequently need to cluster projects based on their scope similarity to support various construction planning tasks such as cost estimation. Few recent studies have presented systematic methods that employ cost composition and pay item descriptions for automated project clustering. However, they suffer from two main drawbacks, including the reliance on unit bid prices, which are unavailable at the time cost estimation is conducted, and the lack of thorough validation of their effectiveness in supporting cost estimation. To address these limitations, this study presents a novel quantity-weighted term frequency–inverse document frequency (QW-TFIDF) project vectorization method using both the text description and quantity information of pay items. QW-TFIDF was validated in terms of its effectiveness in supporting project clustering and cost estimating. Its performance was compared with state-of-the-art approaches, including cost-weighted term frequency–inverse document frequency (CW-TF-IDF) and pay item cost-based similarity determination methods. The results showcased the superiority of the new method over existing ones, thus providing a new means for SHAs to enhance their project clustering practices, particularly in early-stage cost estimation, which in turn, will facilitate better budget forecasting, cost management, and resource allocation. | |
