Differential Privacy in Consumer Behavior Analysis

Consumer behavior analysis is the key enabler for many industrial applications. This is also true for the electricity sector. However, such analysis is based on huge amount of data, which raises the public concern over private information leakage. In this paper, we seek to understand how privacy preserving mechanism may affect the behavior analysis performance. Specifically, we use k-means clustering as an example of behavior analysis and define cluster stability from a probability theoretic viewpoint. We establish the relationship between different levels of privacy preserving requirement and cluster stability theoretically and empirically. Numerical studies highlight the value of our proposed analysis for both system operator and the consumer.