Survey-based uncertainty assessment and FEINN prediction in Indian construction projects

The Indian construction industry plays an important role in economic growth but faces delays, cost overruns, and performance inefficiencies due to uncertainties in material availability, labour, regulations, and environmental conditions. This manuscript presents a survey-based assessment of key uncertainty factors affecting construction project performance in India and develops a predictive model based on the identified factors. A structured questionnaire, developed from an extensive literature review, was administered to 50 engineers at a leading Indian construction firm, achieving a 90% response rate. Exploratory factor analysis (EFA), Kaiser–Meyer–Olkin, and Bartlett’s test were employed to identify and validate the major uncertainty dimensions. The novelty lies in integrating statistically validated uncertainty dimensions extracted through EFA with a finite element interpolated neural network (FEINN)-based predictive framework, combining interpretable survey analytics with non-linear project performance prediction. The research uncovers major uncertainty dimensions including financial instability, labour shortages, material availability, and regulatory delays. The FEINN model achieved a lower prediction error with a root mean squared error of 0.12 and mean absolute error of 0.07, and a mean absolute percentage error of 2.7%, outperforming benchmarks. This approach offers valuable insights to enhance decision making and improve project outcomes across the Indian construction sector.