Time Series Decomposition for Anomalous E-commerce Transactions
Time Series Decomposition for Anomalous E-commerce Transactions

Summary/Abstract: Online trading is one of the pillars of the digital economy. The rapid increase of e-commerce transactions has increased the risk exposure of providers and made it virtually impossible to track consumer behavior by relying on human experts alone. Here we show how time series decomposition can be used to automatically detect suspicious transactions and flag them out for subsequent actions. The identified outliers have clear business meaning and can be interpreted as peaks in demand produced by idiosyncratic consumer behavior or by malicious activity. Either way, they deserve sufficient attention and active management.

• Details
• Contents