This paper aims to analyze optimal resolution for energy consumption data to improve data-driven building energy consumption diagnosis with the ASHRAE change point model. A 9-floor office building is selected as a target, and hourly data is measured throughout a year. Heating, cooling, and other types of energy consumption are measured separately, to find truth values of the heating change point and cooling change point, which refers to the temperature at which the building starts heating or cooling. After the truth change point is found, the total sum of the consumption is resampled along different time intervals, to reflect various levels of data resolution. Change point models are built with each differently resampled data, and their results including change point and RMSE were compared to find the optimal time interval of the data. The result showed 2-week interval is the optimal interval. However, an unordinary pattern, which may be due to individual heating or cooling appliances being counted as appliance consumption rather than cooling or heating consumption, is observed as well. The results of this study suggest that the resolution of public data needs to be improved for the energy consumption diagnosis based on public data, and the usage of individual heating and cooling devices needs to be considered additionally in the diagnosis process.