Mining group correlations over data streams

Mining correlation over steams attracts a lot of attentions recently. However, group correlation analysis over data streams is relatively few. Moreover, existing literatures are mainly focused on a single time window, with large space and time complexity. This paper proposes an online canonical correlation analysis algorithm called MGDS (Mining Group Data Streams). Based on base-window, the MGDS algorithm dynamically maintains a few statistics from raw data to calculate correlation. The mining range is not limited in a single window, but can be changed according to queries. Theoretical analysis and experimental results show that the algorithm is accurate and efficient with space and time overhead reduced greatly.