Detailed simulation studies of building performance can result in large data sets, particularly where statistical information on annual energy or environmental performance is required. Key performance indicators such as the number of hours above a certain temperature can easily be extracted. However, it is difficult for users to explore such datasets and understand the underlying reasons why a building performs in a certain way. This is especially true in climate responsive buildings which involve complex interactions of ventilation, solar gains, internal gains and thermal mass, for example. Data mining techniques have traditionally been employed in the financial and marketing sectors to elicit patterns within the data. This paper describes how the different data mining techniques may be employed in helping to analyse building performance data. Clustering is identified as a particular useful analysis technique and its potential is illustrated through a number of case studies.
|Number of pages||14|
|Journal||Building Services Engineering Research and Technology|
|Publication status||Published - 2004|
- data mining
- building simulation