Figure 4
The flowchart begins with an input layer that includes a global dataset of 103 delay risk factors from a systematic literature review and data extraction from international studies. This input is filtered by applying the top 40% relevance filter to narrow down the factors to 55 key risks. The filtered risks then undergo expert validation and optimisation through semi-structured interviews with 11 Australian construction experts. Kendall's W is computed to measure consensus and ranking stability. A genetic algorithm is applied to optimise expert subset weighting. The final output layer presents a localised set of 22 delay risk factors relevant to Australian construction projects, which are empirically validated and context-specific.Conceptual model of the risk localisation process. Source: Authors’ own work