Towards developing a predictive model for interpersonal communication quality in construction projects: An ensemble artificial intelligence-based approach

Rahimian, Ali; Sadeghzadeh, Keivan; Mohandes, Saeed Reza; Martek, Igor; Manu, Patrick; Antwi-Afari, Maxwell Fordjour; Mirvalad, Sajjad; Odeh, Ibrahim

Abstract

High-quality communication among stakeholders is salient to securing and maintaining collaboration in construction projects. Indeed, the absence of such communication among the workforce leads to inefficiency, low productivity, and substandard deliverables. Against this backdrop, the body of relevant knowledge is bereft of a study investigating the association between workers' interpersonal skills and interpersonal communication (IC) quality. Thus, this study aims to predict the quality of professionals' IC through a multi-pronged artificial intelligence-based methodological approach. In doing so, the literature is reviewed to capture noticable interpersonal skills (IPSs), followed by utilizing a fuzzy-based algorithm to prioritize them. Then, an extreme Gradient Boosting (XGBoost)-based algorithm is developed to predict the quality of workers' IC. The developed XGBoost is finally applied to three real-life construction projects to check its efficacy. Based on the application of the developed model to the selected case studies, the following conclusions are drawn: (1) the significant skills are "Leadership Style," "Listening," "Team Building," and "Clarifying Expectations"; and (2) the predictions of the developed model equal to what happens to the workers' IC quality in more than 78% of the cases. The developed algorithm can warn interpersonal conflicts before they escalate, enhance job-site productivity, team development, and human resources management, and guide construction managers in developing IPSs training.

• Journal Article Type: Article
• Acceptance Date: May 16, 2024
• Online Publication Date: Feb 14, 2025
• Deposit Date: May 17, 2024
• Publicly Available Date: Mar 15, 2025
• Print ISSN: 0969-9988
• Electronic ISSN: 1365-232X
• Publisher: Emerald
• Peer Reviewed: Peer Reviewed
• DOI: https://doi.org/10.1108/ECAM-09-2023-0958
• Public URL: https://uwe-repository.worktribe.com/output/11999238