Energy Performance for Buildings Evaluated with Multivariate Analysis
2005 (English)In: Proceedings of the of the Ninth IBPSA Conference and Exhibition, Building Simulation 2005, 2005, 891-898 p.Conference paper (Refereed)
Numerous problems can occur for an investigatorof larger datasets, e.g. how to handledimensionality, many variables and fewobservations, few variables and many observations,correlations, missing data, noise and to extractinformation from all data simultaneously.Multivariate analysis (MVA) is an establishedmethod for dealing with such problems.In this work, we introduce a methodology based onMVA, which was developed to model the buildingenergy performance from the perspective of theproperty holder. Data from a Swedish database of500 buildings, which recently has been compiledand is under expansion, was used for theinvestigation. The available data consists ofbuilding specific information and consumptiondata, monitored on a monthly basis, reported by theproperty holder. Electrical consumption for lightingand appliances is paid by the tenants in Sweden,and is thus lacking in the database. This means thatthe data base just include the part of the total energyuse that is paid by the property holder. With theoverall goal to assess the energy use paid by theproperty holders, a methodology is suggested forestimating the electrical energy paid by the tenants.At this early stage of our work, we found that theused methodology gives a fairly robust model andthat the interpretation of the model is believed to beaccurate in terms of comparing the energy usebetween different buildings.
Place, publisher, year, edition, pages
2005. 891-898 p.
IdentifiersURN: urn:nbn:se:umu:diva-20204OAI: oai:DiVA.org:umu-20204DiVA: diva2:208243