Abstract: Suspicious transactions are hidden in thousands of massive transaction data, causing incalculable losses and risks, but the detection is very difficult. In terms of how to effectively explore and identify suspicious transactions from massive transaction data accurately and quickly, this paper adopts the method based on the combination of k-means algorithm and random forest algorithm to solve the problem of data imbalance in the identification of suspicious transactions in bank accounts, and proposes an effective suspicious transaction detection model. At the same time, the AUC(Area Under Curve) and Recall indicators such as the unbalanced data classification standard of performance evaluation, and finally to Kaggle data platform for the bank account of suspicious transactions data set, the results show that the proposed detection model of performance evaluation index AUC increased by 5%, F1-measure increases by 1%, show that the method has some reference value to the suspicious transactions recognition, limited information utilization rate is higher, which makes all kinds of Banks prediction speed and accuracy of suspicious transactions events get improved, can to some extent, reduce the operating cost and risk of the banking sector.

Keywords: Big data, K-means Algorithm, Random Forest, Suspicious Identification, Unbalanced Data