 This paper proposes a new method for disaggregating energy consumption data collected from smart meters into different types of loads. The method uses low frequency meter reading and constructs a multi-dimensional feature space with adaptations to smart meter parameters. The method is then used to identify type I and type II loads, as well as timed loads. The algorithm is tested using two data sets, one private and one public, and shows promising results. This article was authored by Amiti Kligman, Arbel Yanov, and Yuval Beck.