Predicting the Occurrence of Construction Disputes Using Machine Learning Techniques

The construction industry is overwhelmed by an increasing number and severity of disputes. primary objective this research to predict the occurrence disputes utilizing machine learning (ML) techniques on empirical data. For reason, variables affecting dispute were identified from literature, a conceptual model was developed depict common factors. Based model, questionnaire designed collect data experts. Chi-square tests conducted reveal associations between input occurrence. Alternative classification tested, support vector (SVM) classifiers achieved best average accuracy (90.46%). Ensemble combining tested for enhanced prediction performance. Experimental results showed that ensemble classifier, obtained majority voting technique, can achieve 91.11% accuracy. tests, most influential factors found as variations unexpected events in projects. Other important predictors all related skills parties involved. This study contributes domain three ways: (1) proposing combined diverse efforts literature identifying occurrence; (2) highlighting factors, such response rate communication skills, indicators potential disputes; (3) providing ML-based with capabilities function early-warning mechanism decision-makers.