With a hot summer, cold winter and high humidity climate, residential energy consumption in the Yangtze River Basin is strongly affected by diverse air-conditioning behaviors in such a harsh indoor thermal environment. The development of big data technology provides a basis for larger samples, higher accuracy, and more dimensions of air-conditioning behavior monitoring, which can make up for the current situation of large errors in existing research methods and single classification indicators. By selecting 2 000 samples of residential room air conditioners (RACs) in Chongqing as the representative city: First, five characteristic parameters of air-conditioning operation are constructed from the perspective of air-conditioning using period, temperature demand and energy consumption; Then, a multi-dimensional clustering algorithm was used to identify the typical categories of air-conditioning behavior;Finally,through in-depth analysis of the characteristic differences among the clustering results, three typical air-conditioning behavior groups are summarized for residential buildings in Chongqing.