Q-statistics showed that electricity energy use data have more significant errors.
Totally different curves appear for train year 2012 and test year 2010.
Faulty data are appeared until June 2011, and after that, data shows increase in quality.
Energy use model was made for heating energy use, considering that data for electricity energy use should be previously corrected. PCR and PLSR models were created. PCR explains more of model variance, while using the same number of components as PLSR model. Modeling with PCR with one Principal Component explains more 99.5% of model, while PLSR explains only 62%. With only four Principal Components PCR model explains close to 100% of model. By using procedure for model scaling and finding driving variables based on PLS weights, it was found that the most important variables of the heating energy use are outside walls area, windows, doors and glass area, and heating volume. Q-statistics shows that highest error is calculated for building 360 Realfabygget (Natural science building).