Effect of Data Repair on Mining Network Streams

Data quality issues have special implications in network data. Data glitches are propagated rapidly along pathways dictated by the hierarchy and topology of the network. In this paper, we use temporal data from a vast data network to study data glitches and their effect on network monitoring tasks such as anomaly detection. We demonstrate the consequences of cleaning the data, and develop targeted and customized cleaning strategies by exploiting the network hierarchy.