Current electricity energy rate is based on a progressive rate with three usage ranges. For an efficient demand management, the time-of-unit (TOU) rate has been developed. In this paper, we employ a support vector machine (SVM) technique to provide a reliable guideline of selecting an appropriate electricity rate for a given household. Here, we assume that only monthly electricity usage and bill of the current progressive rate are available. In order to design a classifier for the selection, we first train two classifiers based on SVM by using two different house complexes, respectively. This separate training can be helpful to check heterogeneous statistical properties of the two complexes. In this paper, we perform preliminary analysis to investigate such different properties, which can be useful in constructing a robust classifier as a future work.