Financial Data Anomaly Detection Based on Data Mining Technology
Fen Song · 2024
In order to understand the anomaly detection of financial data, this paper proposes a research method for anomaly transaction detection based on data mining, which can detect anomalies in transactions at both the business and operational levels. Firstly, when a user submits a new consumption transaction, this paper uses Bayesian belief network algorithm to determine the posterior probability that the current transaction belongs to a normal transaction, as the trust factor at the business level; Then extract several operations of the user before the current transaction, and together with the current transaction, form a fixed length operation sequence. Use the BLAST-SSAHA algorithm to compare it with the user's normal operation sequence and known abnormal operation sequence, and obtain the credibility factor at the operation level. Taking into account both the credibility factor at the business level and the credibility factor at the operational level, the final decision is made on whether the current transaction is an abnormal transaction.