The volume of annual, monthly, and hourly simulation output developed by building simulation packages such as DOE-2.1 presents the building modeler with significant challenges. Developing hourly total load and end-use estimates of building performance calibrated to 15 minute or hourly metered total load or end use data requires new analytic tools that allow the modeler to quickly review the results and make iterative changes to the models. This paper suggests that engineering model calibration can quickly go beyond monthly customer billing data to minimize self-canceling errors where under-predictions in one end-use cancel out: overpredictions in another end-use. New visual data analysis techniques (VDA) can provide the modeler with tools that improve the accuracy of hourly total load and end end-use data, while minimizing modeling costs. This paper presents four calibration approaches and supporting statistical measures that can be used as iterative diagnostic tools for the building modeler.