Data for building energy (LL84)  is processed to remove empty data (NaN) and anomalous outliers to prepare it for an appropriate model of performance. LL84 and PLUTO data sets are merged on the common category, Building, Block, Lot (BBL) a common classifier to then be analyzed and explored for models to benchmark the current state of MF performance.  The number of buildings are counted to show the most frequent is the multi-family building type (Fig. 1).  The data was plotted to view the distribution. A mean of 87 kbtu/SF is found and focus on the poorest performing quartile where EUI is over 100 kbtu/SF (Fig. 2).