Clustering and Machine Learning‐Based Risk Prioritization for Irrigation Construction Projects in Thailand

ABSTRACT Thailand's construction sector has expanded steadily, yet irrigation projects continue to face delays, cost overruns and disputes rooted in inadequate risk management. This study identifies, prioritizes and analyses critical risks affecting irrigation construction performance. Survey data from 460 respondents were evaluated using a probability–impact matrix to derive risk scores and then augmented with machine‐learning technique clustering, correlation heat maps and predictive modelling. Five principal risks emerged: (1) inadequate and unclear owner planning, (2) owner error in issuing work suspensions, (3) delays in material approval, (4) contractor negligence regarding safety and (5) contractor breach of contract. Clustering revealed heterogeneous stakeholder risk perceptions, while heat maps highlighted strong interdependencies among owner‐related risks. Logistic regression and random forest models achieved near‐perfect classification of high‐ versus low‐risk projects, underscoring the predictive value of survey‐based indicators. The findings recommend strengthening early‐stage planning, streamlining approval processes and enforcing contractor accountability and demonstrate how integrating quantitative scoring with data‐driven analytics advances risk assessment and supports sustainable irrigation infrastructure development in Thailand.