An Artificial Intelligence Approach to Establishing Critical Factors Influencing Offsite Construction Adoption in The United Kingdom

Offsite Construction (OSC) is a construction prefabrication concept that is being rejuvenated. Scholars have evaluated factors affecting the wider uptake of the re-introduced product, but adoption challenges persist despite acclaimed benefits. Faced with increasing housing deficit, environmental sustainability and the ambitious 2050 net-zero target, the United Kingdom (UK) government eagerly seeks ways to solve the OSC adoption puzzle because of its potential in addressing these challenges or stop further investment of tax-payers money in OSC innovation as some UK Parliament members suggested. This study, therefore, took a more rigorous approach to evaluating the factors influencing OSC adoption in the UK, relying on a mixed method approach. It aims to establish the critical OSC adoption influencing factors in the United Kingdom and to develop a predictive model using machine learning techniques for predicting adoption likelihood based on the identified critical factors. This study appraised existing OSC factors through extant literature using systematic literature reviews (SLRs). The first review explored the various factors influencing OSC globally. The second review highlighted analytical approaches used to identify the factors influencing OSC adoption and the theoretical approach used to address the article’s research aim. The third review evaluated the state of OSC development in the UK through the lens of the Royal Institute of British Architects Plan of Work (RIBA-PoW). The study was hinged on three theoretical underpinnings, namely the Concern-Based Adoption Model, ADKAR change model and the RIBA-PoW. Employing pragmatism as its philosophical stance and using a mixed method research approach, findings from the first SLR were used to interviews twelve UK construction industry (UKCI) professionals. The SLRs and interview data were analysed using content and thematic analyses, including appraising the findings under the lens of RIBA-PoW. The study triangulated findings from SLR 3 and interview analysis to develop a survey questionnaire used to quantitatively evaluate OSC influencing factors identified. The survey targeted 500 responses; however, 337 responses were received. 315 responses from the 337 were valid for quantitative analysis and predictive model development. The quantitative approach used exploratory analysis in setting the tone, before conducting a correlation analysis to drop features with multicollinearity effect on the data. Employing multiclass classification and binary classification and using 10 feature selection methods, the analysis identified eight very critical factors that show tendencies to influence OSC adoption significantly. They are Awareness of Planning Authority officials (AW4), Design lead timethe amount of time required before freezing OSC design (DCF1), Scalability- Repeatability of design and project modules (DCF10) Availability of past OSC project data to support use of OSC for proposed projects (PKD2), Monotonous perception about OSC designs (PP5), Road width that fits OSC delivery trucks (SCL7), Access Road to the site that fits truck size (SCL8) and Site Hedging (SCL11) . The study developed an adoption model for possible prediction of OSC adoption in the UK following exploratory data analysis and feature selection done. Logistic Regression, LR, (Random State 42) employing Mode imputation for Significant features, in a multi-class experiment using a splitting ratio 80:20, having an accuracy score of 85.71% emerged as the best predictive model, displaying no form of overfitting across all ML development measures. The research contributed to theory, practise and policy. The study advances theory by providing a structured interpretive lens to understand behavioural and organisational readiness that can redefine the approach to unearthing OSC adoption issues beyond what extant literatures present. Additionally, the study adopted methodological and analytical pluralism, integrating literature synthesis, qualitative triangulation, and ML-based predictive modelling -a rare but powerful analytical combination in OSC adoption studies. Further, the study extended ADKAR model to ADKAR⇌A. By extending ADKAR model to ADKAR⇌A, the study argued that showcasing successful OSC projects creates awareness (A)–– a means to increasing OSC knowledge and consequently, potential adoption. Practical-wise, the study developed a framework for aggregating OSC project data for OSC adoption prediction in the UK- a template that can be adopted globally. The ML model developed is the first OSC adoption predictive model for the pursuit of wider OSC adoption in the UK- a model that provides significant generalisable and explainable AI approach to OSC adoption, irrespective of the category or type of OSC, if well implemented. Environmentally, this OSC research findings significantly contribute to cleaner construction production and environment, and they promote Sustainable Development Goals (SDGs) 9, 11 and 13 in pursuit of a net-zero carbon target. Policy-wise, the study’s findings illuminate infrastructural, behavioural, technical, and policylevels levers that can be targeted to boost OSC uptake. The study approach and finding offers the formulation of targeted adoption strategies for academics, industry stakeholders and policymakers and focal points for interventions. The findings also provide better theoretical understanding of OSC adoption factors to help UKCI concert adoption pursuit efforts rightly and the government, to appropriate more pragmatic policies in achieving the targeted 25% OSC usage in construction projects.