Data processing method of residential water and gas data in northern cold zone based on data mining technology

The amount of residential water and gas data system for a city is too large to be manually processed, which requires the support of data mining technology. Based on a 2-year survey on the water and gas data of three communities in Tianjin, presents the processes and results of data processing, analyses outlier detection of energy use and its change in the adjacent two months and compares the overall energy use levels among the three communities, using data mining methods such as data normalization, outlier detection based on proximity and boxplot. Combined with the questionnaire survey data, proposes a data mining approach to explore the correlation between occupants’ energy use levels and their social characteristics and energy related behaviors through information gain theory and C4.5 decision tree. Presents the methodology of extracting useful information from building energy use data, which is expected to assist the platform construction of energy use data management and its application.