The rapidly growing and gigantic body of stored data in the building field, coupled with the need for data
analysis, has generated an urgent need for powerful tools that can extract hidden but useful knowledge
of building performance improvement from large data sets. As an emerging subfield of computer science,
data mining technologies suit this need well and have been proposed for relevant knowledge discovery
in the past several years. Aimed to highlight recent advances, this paper provides an overview of the
studies undertaking the two main data mining tasks (i.e. predictive tasks and descriptive tasks) in the
building field. Based on the overview, major challenges and future research trends are also discussed.