In South Korea, about 40,000 buildings of low-income households have been diagnosed and remodeled annually by the Energy Welfare Program, using the normative method. The normative method is based on the heat gain elements of a building. In contrast, the performance-based method is based on the output derived from the thermal performance of each building part. In the normative method, there is no other building in which the input conditions of a building are perfectly matched. Further, cost-effective energy remodeling strategies vary according to the capacity of the diagnosis engineer. In this paper, we analyze the thermal performance of buildings by the performance-based method using indoor temperature, and examine the possibility of a database for the optimal remodeling method. For this, we analyzed more than 2500 simulation cases by combining thermal performance of each building part. The indoor temperature pattern can be similar even when the thermal performance of each building part is different. In buildings with similar indoor temperature patterns, the coefficient of variation of the root mean squared error of energy demand falls within the acceptable error range. Furthermore, changes in energy demand and predicted mean vote are similar when window thermal performance is changed.