Flexible Data Mining for Prioritizing Critical Building Components

Abstract

Prompt diagnosis and resolution of building problems is essential for facility management personnel to maintain an acceptable level of service for their buildings. While the condition of many building components is assessed based on detailed inspection reports, some components depend more on maintenance requests received from occupants. Currently, facility management teams analyze these variety of data and decide on the required rehabilitation strategy, which is a manual process that is inefficient and subjective. This paper demonstrates the flexible use of data mining techniques on two sample building components with different data types, to support asset rehabilitation decisions. The paper focuses on roofing elements as an example of hidden building systems that are assessed by expert inspectors, and HVAC systems which are assessed based mainly on occupants’ feedback. The paper discusses how data mining can be applied flexibly to data samples of the two building components. The presented approach aims to streamline the asset prioritization process for large owner organizations such as university campuses and school boards, especially in the case of limited budgets.