Research on the Application of Frequent Item Mining in Credit Risks

Abstract

Frequent item mining is the mining of data items that appear at higher frequencies than a designated threshold during data transmission. So far, a more mature algorithm for frequent item mining is finding Frequent Items in data streams using Extensible and Scalable Bloom Filter based on Landmark window model (FI-ESBFL), which deeply explores the essentials of data stream frequent item mining algorithm over decaying window and frequent item judgment based on sliding window model. Based on FI-ESBFL frequent item mining algorithm, this paper further determines the threshold of credit risks with historic sample sets and maintenance sample sets, using the summary information in credit risks, achieves relatively accurate threshold precision with certain storage spaces and provides theoretical guidance for risk management.