Introduction
In the context of building performance, the discrepancy between building energy simulation and measured observations is the focus of many studies \cite{Menezes_2012,de_Wilde_2014,Attia_2013,Fedoruk_2015}.
Since building energy simulation models deal with an extensive number of input and parameters, imprecision and lack of knowledge is inevitable. These short–comes, also known as uncertainties, are major contributors to the gap between simulated and measured building energy performance.
However, it is commonly believed that obtaining a correct understanding of imprecisions as well as increasing the knowledge in unknown contributors can result in better predictions. In order to achieve a better understanding of the uncertainties and monitor their effects, various techniques are invented. These methods span from the initial part of collecting data for simulation input to post processing each and every effect of the input parameters. Uncertainty Analysis (UA) deals with quantitative modeling, which cannot be associated with certainty \cite{Kurowicka_2006}.The approach has various applications in risk assessment\cite{Jiang_2013}, reliability assessment\cite{Zhang_2013} and decision-making (decision support)\cite{Kim_2014}.
The uncertainty in Building Energy Simulation (BES) can be addressed considering various sources. Hopfe and Hensen \cite{Hopfe_2011} assumed three sources: uncertainty in physical, design and scenario
parameters. They focused on the physical properties of wall, roof and
floor area as physical uncertainties. The uncertainty in design
parameters was referred to different design strategies, which are
adopted in the planning process. Finally, the scenario uncertainty that
is derived the future life of a building. The occupancy and future
weather condition are the main causes of these kinds of uncertainties.
Fig. 1 displays the different sources of uncertainty and their
subcategories in BES.