Financial object single abnormal attribute analysis method based on association analysis

Abstract

The invention discloses a financial object single abnormal attribute analysis method based on association analysis, and the method comprises the steps: S1, selecting a factor, obtaining different object information under the factor, and dividing to-be-screened financial transaction data to each object; S2, calculating a transaction failure rate and a failure contribution degree under each object;S3, establishing a projection coordinate system; S4, screening out abnormal objects; S5, selecting a target financial object, and taking given remaining factors except the selected factors as attributes of the target financial object; S6, obtaining a single abnormal attribute set A of the target financial object; S7, analyzing a single abnormal attribute set B of the target financial object; S8, combining the single abnormal attribute sets A and B to obtain a union set D; and S9, for different abnormal objects, repeatedly executing the step S5 and the step S8 to obtain a single abnormal attribute set under each abnormal object. An accurate single abnormal attribute set can be obtained, and the reason of transaction failure can be quickly and completely determined.

Description

technical field [0001] The present invention relates to financial data analysis, in particular to a method for analyzing single abnormal attributes of financial objects based on correlation analysis. Background technique [0002] As far as today's society is concerned, mobile payment has become the mainstream. How to submit transaction success rate and enhance user experience is a problem that all payment companies, related merchants and banking institutions need to pay attention to. Payment is a long-link transaction activity with many participants, including: cardholders or payment users, merchants, acquirers, card issuers, Nets Union or UnionPay and other related institutions. The factor that causes the transaction failure may be a single factor, but It is more a combination or superposition of multiple factors, and many failure factors cannot be judged simply by one point. In addition, the success rate of current transactions is very high. Generally, the overall transac...

AI Technical Summary

Problems solved by technology

Payment is a long-link transaction activity with many participants, including: cardholders or payment users, merchants, acquirers, card issuers, Nets Union or UnionPay and other related institutions. The factor that causes the transaction failure may be a single factor, but It is more a combination or superposition of multiple factors, and many failure factors cannot be judged simply by one point

In addition, the current transaction success rate is very high, and the overall transaction success rate is generally around 99%. Facing massive payment transaction data, how to find and discover the problems and reasons for transaction failure is very difficult.

[0003] For the data of failed transactions, the cause of failure can be analyzed by the proportion of failed data in different objects or different attributes, but payment failure transaction data has several obvious characteristics: 1. Massive data; 2. Data skew is very serious ( The proportion of failed transactions is very small); 3. The data of small institutions or merchants (including certain types of business) is extremely easy to be overwhelmed, for example: An institution has a high transaction success rate of 98%, but a certain type of new business fails due to some reasons However, the rate is very high, and almost all of them fail. It is difficult to find such small-volume transactions on the whole

[0004] At the same time, for the failed transaction data, the failure response code fed back by the financial system can also be used to identify the cause of the transaction failure. However, some failure response codes currently involve multiple attributes. In this case, it is impossible to obtain a clear analysis result. Facilitates rapid determination of abnormal properties

Images