A digital framework about determining rates of inherent defects insurance based on machine learning

Inherent defects insurance (IDI) plays a critical role in mitigating quality risks and enhancing construction standards in construction engineering. However, the determination process on IDI rate lacks intelligent data analysis and differentiation in rate outcomes, with limited digitalisation and automation in rate-setting processes. This has led to low engagement from developers, and as a result, IDI has yet to be widely adopted, failing to realise its full potential. This paper combines case-based reasoning and random forest (RF) algorithms to build a digitalised IDI rate-setting framework. The Analytic Hierarchy Process is applied for subjective weighting of attributes to analyse key factors influencing IDI rates and 17 IDI cases are collected and then digitised into tuple-based forms to build a case library. By integrating RF, the paper calculates the similarity between cases based on the weighted attributes to perform case retrieval. In addition, RF is used to predict rates intelligently for target cases that do not have retrieval results. This digital framework enables targeted determination of IDI rates based on data analysis, supports management personnel in making decisions, and promotes the effective implementation of IDI.