 clustering techniques are commonly used to group similar objects together from a large data set. However, when dealing with correlated data, such as time series, it can be difficult to determine which similarity measure should be used. This paper proposes a new approach to validate the performance of different clustering algorithms by comparing their results using a balanced vector score. Additionally, this paper also explores the use of clustering techniques for forecasting and modeling building energy consumption patterns. This article was authored by Wolfgang Kastner and Felix Iglesias.